Inflation targeting, policy rates and exchange rate volatility: evidence from Turkey.

Akyurek, Cem ; Kutan, Ali M.

INTRODUCTION

Facing a sudden reversal of capital inflows, Turkey had to exit from its 13-month-old exchange rate peg and adopt a floating exchange rate regime in February 2001. The after effects were devastating, as the nominal exchange rate nearly doubled in 2 months, inflation more than doubled in 6 months and the economy contracted significantly with the collapse of the banking system. The crisis led to a new economic programme funded by the IMF and a new approach to monetary policy (Akyurek, 2006). Accordingly, in January 2002 the Central Bank of Turkey (CBT) began implementing an implicit inflation targeting framework (IIT), which encompassed core attributes of an inflation targeting (IT) regime including, among other requirements, the announcement of a formal target for inflation. Coupled with fiscal discipline and economic reforms, under the IIT framework, inflation declined from over 70% to below 8% by year-end 2005 while the CBT consistently outperformed its year-end point targets measured by the consumer price index (CPI). In due course, the CBT cut its overnight policy rate from 59% to 13.5%. It seemed more or less that during this period of disinflation, structural transformation and transitional dynamics, the CBT was searching for an equilibrium real interest rate commensurate with lower inflation.

Given successful disinflation, improvements in fiscal conditions and the banking system in particular, the CBT introduced full-fledged IT at the beginning of 2006, which brought further transparency to the new framework. During the first year of full-fledged IT, year-end inflation (9.86%) exceeded the target (5%) by a wide margin. The year saw a sudden reversal of capital inflows in May-June mainly driven by the deterioration in investor sentiment towards emerging markets in general. The exchange rate depreciated by close to 30% and the CBT raised its policy rate from 13.25% to 17.50%. While inflation returned to pre-shock levels after about a year, it remained above the path leading to the 4% year-end target set for 2007.

In this paper, we focus on Turkey's inflation targeting regime. Turkey's experience is important for several reasons. First, literature on inflation targeting mainly focuses on developed countries. (1) More recently, some developing and emerging markets have adopted a regime of inflation targeting and their experience is examined by some recent studies. (2) Many new member states of the European Union (EU) have also adopted inflation targeting regimes. (3) However, there is scant work on inflation targeting regime for Turkey. We believe that Turkey's experience may provide valuable lessons for emerging European and other developing economies. In addition, Turkey is a candidate for the EU membership, and inflation targeting is consistent with the Maastricht criteria for joining the European Monetary Union.

The paper has three simple objectives. Firstly, in the second section, we provide a short description of the IIT period in Turkey including a discussion of the evolution of the new monetary framework and the conditions prevailing during the period of its implementation. Here we emphasise the challenges facing the conduct of monetary policy in this new environment and also discuss the changes to the framework with the transition from IIT to IT. The third section provides a discussion of theoretical issues concerning interest rate policy in small open economies under the IT regime. The section provides a brief discussion of monetary transmission with an emphasis on issues concerning the dynamics of monetary channels in emerging markets and implications for monetary rules. The fourth section provides a discussion and empirical analysis of the path of policy rates in Turkey where the main drivers of policy changes seen under the new framework are formally investigated. As such, a Taylor rule is estimated for 2002-2007. In the section Exchange rate pass-through, capital inflows and monetary policy: 2002-2007, the results of an empirical analysis of exchange rate pass-through to inflation is presented, including an investigation of changes, if any, in the process since the introduction of IIT. The next subsection includes a brief discussion of monetary policy reaction to sudden stops during 2002-2007. Concluding remarks are presented in the last section.

THE NEW MONETARY POLICY FRAMEWORK AND PRELIMINARY CHALLENGES

The literature provides a long list of requirements that countries should meet if their IT framework is to operate successfully. (4) Some are structural requirements and seem more IT-specific when taken as indispensable parts of the framework. These include (i) public announcement of medium-term inflation targets; (ii) institutional commitment to price stability as the primary goal of monetary policy, to which other goals are subordinated; (iii) increased communication with the public and markets about the plans and decisions of the central bank; and (iv) increased accountability of the central bank for attaining its inflation objective. In this context, Mishkin (2000) emphasises the finding that IT entails much more than a public announcement of numerical targets for inflation for the year ahead and that it is particularly important for emerging market countries to establish all other conditions if their IT regimes are to be sustainable over the medium term. The IT regime has the potential to reduce the likelihood that the central bank would fall into the time-inconsistency trap if it increases transparency and the accountability of the central bank, particularly for emerging market countries with a long history of monetary mismanagement and credibility issues. Other requirements often cited for a successful IT regime, which seem as desirable features for successful operation of all available monetary policy frameworks, include (i) a strong fiscal position; (ii) a well-developed financial system; (iii) a reasonably well-developed ability to forecast inflation; and (iv) a well understood transmission mechanism between monetary policy instruments and inflation.

The new monetary framework of Turkey was introduced as part of a comprehensive IMF-supported economic programme following the February 2001 crisis during which the exchange rate peg was replaced by a floating exchange rate. For the balance of 2001, a transition period before the IIT, IMF programme monetary performance criteria were set as quantitative limits on net domestic assets (NDA) and net international reserves (NIR). Yet, an indicative target was set for base money, which the CBT indicated would be the main operational target of monetary policy during 2001. With the commencement of the IIT framework in 2002, the overnight interest rate became the operating target while monetary indicators also remained as performance criteria and indicative targets.

Turkey's IIT regime encompassed many of the core institutional attributes of the full-fledged IT framework. Multi-year point CPI inflation targets were announced and new legislation gave legal independence to the CBT, as achieving and maintaining price stability was declared as its primary objective. The CBT was also fully authorised to choose its monetary policy instrument. Communication with the public, transparency of monetary policy and accountability of the CBT improved significantly with the establishment of a monetary policy committee (MPC), communication of the rationale for policy changes through press releases, the start of regularly published inflation and monetary policy reports and frequent public appearances by CBT officials. Inflation targets were announced as 20%, 12% and 8% for 2002, 2003 and 2004, respectively, based on the CPI index after inflation had declined to 29% at end-2001 for which the official projection was 35%. The advantage of setting the target on broadly defined headline inflation was that it is better understood by the public, while the disadvantage was that its movement could reflect factors other than monetary policy measures, and could be a more difficult target to hit. The CBT preferred to set CPI targets but also relied on narrowly defined indices of inflation, such as those excluding supply shock effects, to assess the path of inflation and to communicate inflation conditions to the public.

The unfavourable economic developments that forced Turkey out of its peg and motivated the shift to an IT-like regime posed risks for its credibility. Recurrence of factors that led to economic crises could force monetary policy to subordinate the newly established inflation objectives. In this respect, there were some key features of the economy in general and characteristics of inflation dynamics in particular were significant obstacles to the success of the new monetary regime at the onset. Turkey had suffered from high and volatile inflation, as average inflation over the past 20 years had registered at 61.5% with a standard deviation of 26%. Inflation had a strong fiscal component, and yet several studies of its dynamics indicated the existence of a strong inertial component, which had progressively strengthened over the years. Additionally, movements in the exchange rate had a profound effect on inflation due to high pass-through from the former to the latter and monetary policy had accommodated several external shocks over the past two decades. (5) The exchange rate-based stabilisation programme introduced in January 2000 failed to achieve the intended results on the inflation front and the forced exit from it resulted in a more than doubling of the nominal exchange rate within a couple of months, which raised inflation to 74% at the beginning of 2002 from 36% at end-2000. Inflation targeting frameworks had not been used to engineer major disinflation from a starting point of high inflation and in this respect the practice of IT in Turkey would be a challenge.

Schaechter et al. (2000) emphasise exchange rate stability as being important for the successful operation of IT regime and, in this context, argue that stability can be defined as a policy framework with an exchange rate value credible enough to convince the markets that the inflation target will not be threatened by currency crisis. The sharp real depreciation in the aftermath of the 2001 crisis and the role IMF support would play to weather pressures on the currency going forward were two factors that provided some comfort regarding stability in the exchange rate at least initially. Yet, past experience with capital inflows and exchange rate volatility was concerning.

As such, capital inflows (outflows) have been the catalyst for booms (busts), as evidenced by the correlation between net capital movements, exchange rate movements and GDP growth over the past several years. Episodic capital inflows relieved the fundamental foreign exchange bottle neck of the economy, cheapening imports of essential intermediate products, and facilitating an expansion of the domestic financial system and bank credit, which led to periodic outbursts in growth. These episodes were unequivocally followed by a sharp weakening in the exchange rate, higher inflation and lower growth.

The fiscal conditions at the start of the IIT framework certainly provided a challenge for monetary policy. Prior to the 2001 crisis, and even more so in the aftermath, the level of public sector debt and concerns regarding its rollover had become the focus of markets. Bonds issued to recapitalise the banking system, the sharp fall in output and rising real rates caused the public sector debt ratio to rise sharply. Additionally, the Treasury issued large sums of foreign exchange denominated and linked bonds as well as floating rate notes, which increased the sensitivity of debt service to fluctuations in short-term interest rates and exchange rates. (6)

The government responded to rising real interest rates following the crisis by doubling the public sector primary surplus (to over 6% in 2001 and 2002), and IMF funding helped significantly to allay rollover concerns. Yet fiscal dominance remained a challenge for the new monetary regime. In the period following the commencement of the IIT regime, there were several episodes of dramatic change in market sentiment and overall expectations of economic agents evidenced by the size and frequency of volatility in domestic bond yields and the nominal exchange rate, the positive covariance of these variables, and abrupt swings in real sector expectations (see Figures 1 and 2). This volatility largely stemmed from changes in market perception of the public sector debt outlook, which depended heavily on the availability or lack thereof of some external support. (7) To the extent that such fluctuations would affect the future path of inflation and growth, they presented serious challenges for monetary policy under a framework that was at its infancy stage. Additionally, there were concerns regarding the sustainability of fiscal discipline, as the turnaround in the primary surplus was largely due to low quality measures and sustainability depended on completion of reforms in challenging areas. (8)

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

Nevertheless, the fiscal conditions gradually improved and became more supportive to monetary policy, as the government performed well on the primary surplus. The debt ratio declined significantly (by 32 percentage points by end-2005), the instrument composition of domestic debt and its maturity improved and the movements in the exchange rate and bond yields decoupled as volatility in markets declined substantially.

The banking sector was weak at the start of IIT and its vulnerability improved gradually. Depreciation of the Turkish Lira, high funding costs and economic contraction eroded the capital base of banks, as non-performing loans mounted. In May 2001, the Banking Sector Regulatory Agency (BRSA) announced a rehabilitation programme and with immediate action public sector banks were re-capitalised, as their accumulated hidden losses were recognised. Among other obvious benefits for banking and the policy environment in general, this was an important step that provided immediate support to the CBT's monetary policy, as public sector banks were heavy net borrowers in the overnight market, which put upward pressure on interest rates. Yet, restructuring of the private banking sector was a gradual process that spread over several months, as recapitalisation was largely left to voluntary capital injection by owners or through mergers carried out by the State Deposit Insurance Fund. (9) The supervisory BRSA role was improved in several ways, which improved risk management and helped to reduce overall risks in the banking sector. (10) Thus, whereas the weakened banking system may have been a challenge for the IIT framework in its early stages, the overall improvement in the sector, particularly the progress in balance sheet risks, gradually eased the concerns of the CBT. (11) The arrival of foreign banks was also a positive development for the banking sector, as between 2002 and mid-2005, foreign ownership of bank shares increased from 3.1% of the total to 12.3%. (12)

Putting all this together, during 2002-2005 the economic environment seemed to have improved progressively for better support of the IIT framework. The improvement in fiscal dominance and provision of official funding together with the higher credibility achieved through outperforming of inflation targets laid the groundwork for an effective monetary policy framework going forward.

At the start of 2006 these positive developments led to the announcement of the full-fledged IT providing further transparency to the monetary policy framework and change in IMF conditionality adding more flexibility to monetary policy.

* The role of the MPC changed from being an 'advisor to the Governor' to 'decision maker'. The targeted index remained as headline CPI and a [+ or -] 2% tolerance band around the inflation path leading to the central target was established. (13) The CBT announced that monetary policy would accommodate the first round effects of inflationary supply shocks (oil price increases, changes in indirect tax rates, etc.) to the extent that forecast inflation remains within the tolerance margin and will react to these shocks only if they lead to second-round effects through deteriorating inflation expectations. Positive shocks to inflation were to be treated in the same manner such that the CBT will react opportunistically and may allow inflation to decline to the bottom of the band while not easing monetary policy.

* Under IIT, the IMF programme conditionality in the monetary area relied on two performance criteria: a ceiling on CBT's NDAs and a floor on its NIR. (14) With full-fledged IT, the NDA target was eliminated. (15) The additional flexibility introduced to the monetary framework through the elimination of NDA performance criteria and base money indicative targets was a significant change to the regime. Given the reverse currency substitution that took place in Turkey during 2002-2005, inflation and base money growth rates decoupled since mid-2003. Thus, while keeping a close eye on developments in monetary aggregates was still important, the increased flexibility to the regime improved the CBT's ability to follow an information inclusive strategy in which many variables could be used for the setting of its overnight interest rate policy tool. (16)

* The changes in 2006 increased the CBT's accountability as well. The CBT began to issue inflation forecasts and in the event of a difference between the inflation outcome and the target, the CBT was obliged to publish a report explaining the underlying causes to the public. Should the CBT project a difference for the period ahead, it would also submit a report to the government explaining the reasons for this variance including a set of measures to be taken to bring inflation back in line with the target path.

OPEN ECONOMY IT REGIME: REVIEW OF SOME POLICY ISSUES FOR EMERGING MARKETS

As demonstrated in Eichengreen (2002), IT monetary rules are generally derived from the following representation of the economy:

[JI.sub.t+1] = [JI.sub.t] + [alpha]]([y.sub.t] - [y.sup.*)] + [[epsilon].sub.t+1] (1)

[y.sub.t+1] - [y.sup.*] = [lambda]([y.sub.t] - [y.sup.*)] - [beta]([r.sub.t] - [r.sup.*)] + [[eta].sub.t+1] (2)

Equation 1 is an accelerationist Phillips Curve, where next period's inflation (JI) is determined by this period's gap between actual output ([y.sub.t]) and potential output ([y.sup.*),] the current inflation level and a disturbance term ([[epsilon].sub.t+1]). Equation 2 is an aggregate demand function where next period's output gap is determined by the deviation of the real interest rate ([r.sub.t]) from its normal level ([r.sup.*)] in the current period, current period's output gap and a disturbance term ([[eta].sub.t+1]).

The key feature of this closed economy representation is that policy rate changes effect inflation through the aggregate demand channel only, which occurs with a lag of two periods. The monetary policy implication of this is that optimal policy under strict inflation targeting calls for setting the inflation target ([JI.sup.*)] two periods ahead, as next period's inflation cannot be influenced by policy rate changes in the current period. Given these assumptions, the optimal central bank reaction function leads to the following well-known Taylor rule, which assigns positive weights on deviations of inflation from target and output from its potential level:

[r.sub.t] = [r.sup.*] + [phi] (JI - JI) + [chi]([y.sub.t] - [y.sup.*)] (3)

where [phi] and [chi] depend on parameters of equations 1 and 2. (17) The parameter

values of the inflation gap and output gap variables would then indicate the degree of flexibility in the IT regime in this closed economy. (18)

Adding the foreign trade component of the economy to the above representation yields the following open economy equations:

[JI.sub.t+1] = [JI.sub.t + [alpha]]([y.sub.t] - [y.sup.*]) - [gamma]([e.sub.t] - [e.sub.t-1] + [[epsilon].sub.t+1] (4)

[y.sub.t+1] - [y.sup.*] = [lambda]([y.sub.t] - [y.sup.*]) - [beta]([r.sub.t] - [r.sup.*]) - [delta][e.sub.t] + [[eta].sub.t+1] (5)

where the exchange rate (e) is defined as the foreign price of domestic currency. (19)

The inclusion of open economy attributes introduces three key features to this model of the economy with significant implications for monetary policy. Firstly, it introduces an additional aggregate demand channel through which policy rate changes may influence. As such, an increase in policy rates will depress net export demand, and thus aggregate demand and inflation, by appreciating the currency. This indirect aggregate demand channel works with a two-period lag as in the closed economy aggregate demand channel. Secondly, this additional aggregate demand channel in the open economy framework increases the response of inflation to policy rate changes. The policy rate adjustment required to close a given inflation gap is smaller the more responsive is the aggregate demand to interest rate changes and the more responsive is the inflation rate to changes in aggregate demand. (20) Thirdly, a direct exchange rate effect on inflation, which takes one period, will occur through a decline in inflation of imported goods as the currency strengthens. Importantly, the emergence of this direct exchange rate channel gives the central bank the ability to influence next period's inflation rate, hence allowing inflation to converge to its target path faster in the event of an inflation gap.

Under the open economy framework, the question of whether IT countries should explicitly consider the exchange rate as a variable in their monetary rule given the existence of these open economy channels is a somewhat unresolved issue. Particularly, how should an IT central bank react to shocks that cause a sharp weakening in the exchange rate? For example how policy rates should be adjusted in response to a sudden reversal of capital inflows resulting from external factors, which weakens the exchange rate temporarily? This transitory shock may require an increase in the policy rate to dampen future demand-induced inflation, yet the increase should target to dampen only the domestic component of inflation not the imported component, which is temporary (Eichengreen, 2002). Likewise Ball (1999) argues that responding to the domestic component of inflation produces better results when shocks to the exchange rate are temporary. The logic is that responding to a problem in the current period that will have disappeared the next period, using an instrument that takes one period to work, will increase inflation and output volatility. Note that this assertion is a direct result of the existence of policy lags such that responding robustly to a temporary shock destabilises the target variables. If the weakening in the exchange rate is the result of a real shock to the foreign component of aggregate demand (to terms of trade or export demand), weaker aggregate demand will be deflationary in the intermediate term. While the correction in the exchange rate will be inflationary, the appropriate response under the IT framework is to allow the exchange rate to adjust and refrain from increasing the interest rate when the economy is weakening (Mishkin, 2000). Eichengreen (2002) on the other hand argues that the central bank may still raise policy rates that will limit depreciation in the short run, while still allowing the exchange rate to adjust eventually to its new long-run equilibrium.

Responding to exchange rate shocks become more complicated for central banks in emerging markets where exchange rate pass-through to inflation is high and liability dollarisation is widespread. Figures 3-5 show the experience of Turkey (see the fifth section for a detailed analysis). In most emerging market economies, given lack of credibility of economic policies and history of high and volatile inflation, economic agent's often fear that transitory shocks to the exchange rate will be accommodated by monetary policy. Thus, with faster exchange rate pass-through the temporary rise in inflation may become permanent. In the framework presented above high pass-through implies that a change in the exchange rate has a larger short-run negative impact on inflation (larger 7 in equation 4) and a smaller short-run positive impact on output (smaller [delta] in equation 5) since nominal depreciations have a weaker impact on the real exchange rate. The implication of this is that it creates a big temptation for the central bank to tighten monetary policy when facing a temporary weakening in the exchange rate (Eichengreen (2002). Because if the central bank tightens monetary policy in response to a weakening in the exchange rate, there is more to gain from disinflation and less to lose in terms of output due to loss of competitiveness. Cabellero and Krishnamurthy (2003) argue that because the principal constraint on output growth in emerging market economies is a shortage of external funding, raising interest rates reduces exchange rate depreciation and hence limits inflation, with limited effects on output beyond the impact of the external constraint.

[FIGURE 3 OMITTED]

[FIGURE 4 OMITTED]

The existence of currency mismatches in emerging markets leaves balance sheets of the private and public sectors vulnerable to exchange rate depreciation. As such, an increase in the exchange rate decreases net worth of the private sector, which is followed by a decline in private sector demand and hence output (Cespedes et at., 2000). Moreover, as the exchange rate depreciation increases the public sector debt ratio in the existence of public sector foreign debt, fiscal policy response in the form of lower expenditure and/or higher taxation also decreases aggregate demand and output. The balance sheet effects also imply a reduction in the value of 3 in equation 5, the conventional positive response of output to currency depreciation. (21)

[FIGURE 5 OMITTED]

The key issue then is that when the exchange rate weakens in emerging market economies, the coexistence of high pass-through and liability dollarisation means that there will be a sharp rise in inflation in the short run but also a strong disinflationary effect in the longer run, that is, post exchange rate pass-through, as indicated by Eichengreen (2002) and Gomez (2004). Carranza et al. (2004) show that inflation may respond negatively to nominal depreciations with this effect being more intense the higher the level of dollarisation of the economy and during the recessionary part of the business cycle. Therefore, in the event of nominal depreciation, an IT central bank should feel less compelled to raise interest rates in order to strengthen the domestic currency and limit the short-term rise in overall inflation due to rising import prices.

Cabellero and Krishnamurthy (2003) argue that the IT central bank will allow the exchange rate to float freely in the face of an external shock; this decreases the demand for imported investment, which in turn decreases the interest rate. Gomez (2004) argues that if central bank monetary policy looks beyond the short term, the central bank is able to decrease interest rates. Whereas if the central bank raises interest rates to minimise the short-term rise in inflation (dubbed fear of floating by Calvo and Reinhart, 2002), monetary policy reaction combined with strong balance sheet effects is likely to be procyclical as the demand for imported investment will increase. This type of reaction by the central bank may be inconsistent with the long run inflation target.

Gomez (2004) also argues that an IT central bank must emphasise that inflation will go back to target after the first round effects of the exchange rate shock. Emphasis on the medium term and the central bank commitment to the long-term target could be the primary elements of the communication strategy under the IT framework. Mishkin (2004) also proposes that under a sharp depreciation, the central bank should focus on inflation after the shock. While depreciation causes inflation in the very short term, monetary policy transmission mechanisms, especially the aggregate demand channel, take about 2 years to effect inflation sizably. In the face of an unanticipated event such as a sudden stop, the central bank is unable to attain the inflation target in the short term. (22)

On the other hand, Ball and Reyes (2004, 2006), Eichengreen (2002) and Mishkin (2004) also argue that with the IT approach to monetary policy; movements in the exchange rate will be taken into account indirectly in setting monetary policy, because the exchange rate affects price behaviour. This will generally produce a pattern of monetary tightening when the exchange rate depreciates, a response similar but not necessarily of the same magnitude, to that which would be undertaken if the exchange rate were being targeted directly. For this reason, an IT country may appear to be showing fear of floating behaviour.

Four main points emerge from these papers. Firstly, in the event of a shock to the exchange rate, the most permanent and thus important effect of monetary policy on inflation and output depends on the central bank's reaction. Secondly, in emerging market economies where exchange rate passthrough is high and liability dollarisation is widespread, the central bank should focus on the domestic component of inflation with the aim of returning inflation to the target path in the longer term when the exchange rate weakens. Thirdly, fear of floating type of reaction by the central bank to an exchange rate shock controls exchange rate pass-through during the first year, but at a cost: a protracted recession in the longer horizon. Finally, there is a fine line between domestic interest rate increases that are associated with exchange rate fluctuations but are due to strict adherence to an open economy IT regime and changes due to fear of floating which imply a hidden policy goal of managing the exchange rate. The problem addressed here is one of measurement.

From a policy perspective, it is also important to note that neither high exchange rate pass-through nor liability dollarisation is exogenous. Exchange rate pass-through is estimated to be higher in emerging market economies. (23) Yet it is also true that adoption of IT has led to lower pass-through in several emerging market economies. (24) Several factors are known to affect the magnitude and speed of pass-through from exchange rate changes to inflation. Among others, the output gap, the type of exchange rate regime in place, the perceived degree of policymakers' commitment to disinflation, trade openness, and the level and volatility of inflation influence the dynamics of the process. The output gap affects the pass-through by reducing the firm's power to increase prices, as during times of increasing sales firms find it easier to pass-through increases in costs to final prices (Goldfajn and Werleng (2000).The inflation level affects the persistence of cost changes, which is positively correlated with pass-through (Taylor, 1993). This view, as expressed by Campa and Goldberg (2002) is that the pass-through of costs into mark-ups is endogenous to a country's inflation performance, generating a virtuous circle where low inflation variability leads to reduced mark ups, less inflationary implications of monetary expansions and continued low mark ups. Credible monetary authorities are expected to act according to the inflation stability objective, which keeps low inflation expectations even in the advent of a large depreciation. Put differently, enhanced credibility of the monetary framework will tame the worries regarding the prospect that transitory exchange rate shocks will be validated and hence become permanent. Output effects of liability dollarisation (currency mismatches) are also not completely independent of the type of monetary regime. As such a flexible exchange rate will encourage hedging by banks and non-banks and discourage currency mismatches. Flexibility in the exchange rate endogenises liability dollarisation and currency mismatches can be reduced to some extent with flexibility and other prudent policies (Goldstein and Turner, 2004). Cabellero and Krishnamurthy (2003) show that fear of floating may seem optimal from a contemporaneous perspective, as it dampens short-term inflation with limited output effects, and yet it is suboptimal ex ante. The reason for this suboptimality is that the anticipation of the central bank's tight monetary policy during the sudden stop has important effects on the private sector's incentives to insure against sudden-stop events. Simply put, contracting dollar debt is less costly in an environment where the domestic currency is expected to be supported in the event of a sudden reversal of capital inflows. Thus, the anticipation of tight monetary policy leaves the economy less insured against the sudden stop. (25)

PATH OF POLICY RATES IN TURKEY DURING 2002-2007: TAYLOR RULE ESTIMATIONS

Four factors in particular seemed to have influenced interest rate policy in the environment of floating exchange rates and IIT. First, the vulnerability of public sector debt required excellent coordination among fiscal policy, public sector debt management and interest rate policy. When markets are concerned about the sustainability of public sector debt, as they were increasingly in Turkey in the aftermath of the 2001 crisis, policy rate changes can have a powerful unconventional impact on expectations. As such, if rates are increased or cuts are delayed, the markets tend to expect future monetisation of debt in which case tighter monetary policy could worsen inflation expectations and lead to higher inflation. Fiscal dominance became lesser of an issue with declining debt ratio, extending maturity and strict adherence to large primary surpluses. Secondly and closely related to the fiscal issue, the perception of risks regarding the adherence to the IMF programme targets and the implementation of structural reforms at times heavily dominated market expectations, particularly in the early stages and seemed to be a key variable affecting policy rate decisions. (26) Thirdly, there were ambiguities regarding the transmission mechanism from interest rates to economic activity and inflation, as indicated by the CBT on several occasions. The effectiveness of the impact of interest rate changes on inflation through the aggregate demand channel was questionable given the relatively low level of financial intermediation. Moreover, the existence of a relatively stable and predictable link between the output gap and inflation was in question as well. Finally, the real exchange rate appreciated strongly (by 38% during 2002-2005), which was due to strong capital inflows and partial de-dollarisation. This complicated the gauging of monetary conditions, as the exchange rate channel of monetary transmission was conjectured to be strong and quick in Turkey. The questions of how much real appreciation the CBT should have allowed and how monetary policy should react to occasional sharp corrections in the exchange rate became important issues shortly after the introduction of IIT.

The CBT's successive rate cuts brought the overnight policy interest rate from 59% to 13.75% during January 2002-April 2006. In the meantime, inflation decreased from 70% to 7.08% by mid-2004 before climbing up to 8.18% in April 2006. In due course, year-end CPI point inflation targets of 35%, 20%, 12% and 8% for 2002, 2003, 2004 and 2005, respectively, were outperformed (see Figure 6). Yet, the 4% target for end-2006 was missed by a wide margin as inflation increased to 9.67% on the back of a temporary weakening in the exchange rate in mid-2006. In reaction to the sudden stop that occurred in May-June 2006, the CBT raised its policy rate to 17.50%.

Eyeballing the data for the entire IT period seems to suggest that the pace of CBT's nominal policy rate changes more or less reflected the pace of decline in inflation expectations, as rate cuts appeared to be more aggressive at times of sharper decline in the credibility gap (inflation expectations less target path) and more passive (smaller cuts or occasional pause) during slow or no convergence of inflation expectations (see Figure 7). The IIT period shows that real rates (forward looking) have remained stationary around a progressively declining plateau (see Figure 8). Real rates were at around 20%-22% until mid-2003, thereafter declining to an average 14% till end-2004 and then to an average slightly below 10% till end-2005 before climbing up to around 12% in the second half of 2006. Yet, a more formal analysis of interest rate policy with the estimation of an open economy Taylor rule provides a simple and easily understood starting point for the analysis of monetary policy during the IT period.

We specify a Taylor rule for the real interest rate (r), dependent on the deviation of expected inflation from the inflation target path (JI-JI*), the output gap (y-y*), the nominal exchange rate depreciation (e) and Turkey's sovereign debt risk premium (EMBITR). (27) The latter is added as a proxy of Turkey's risk premium.

[FIGURE 6 OMITTED]

[FIGURE 7 OMITTED]

[r.sub.t] = [c.sub.1] + [c.sub.2] (JI - JI) + [c.sub.3] (y - [y.sup.*]) + [c.sub.4]e + [c.sub.5]EMBITR + errorterm (6)

[FIGURE 8 OMITTED]

Our sample period is January 2003-December 2006, which is relatively short and therefore the results should be interpreted with some caution. Nevertheless, we believe that the estimations provide statistical facts that help to explain the path of policy rates in Turkey during the IT period (see Table 1). Four points stand out in particular. Firstly, the estimated CBT reaction function tracks the path of policy rates very closely (Figure 9). Secondly, Turkey's risk premium has been a very significant factor determining the path of policy rates during the IT period. Thirdly, results indicate that nominal exchange rate changes and output gap developments have played a relatively minor role in interest rate decisions. (28) Finally, the coefficient of the inflation gap variable is borderline significant.

The significance of the risk premium variable (EMBITR) and the results in general indicate the role of declining equilibrium real rates in CBT's policy function. As further improvements towards building a sustainable fiscal regime and stronger banking sector among other reforms were completed, Turkey's risk premia declined. It seems that during this period of structural transformation and transitional dynamics, the CBT was searching for an equilibrium real rate commensurate with low(er) inflation. The more comfortable the CBT felt with progress on these fronts, the easier and quicker have been the move to lower real rates.

[FIGURE 9 OMITTED]

EXCHANGE RATE PASS-THROUGH, CAPITAL INFLOWS AND MONETARY POLICY: 2002-2007

Exchange rate pass-through: empirical results

Exchange rate movements and inflation expectations have played an important role in inflation dynamics in Turkey over the past several years. (29) The past studies have shown that exchange rate pass-through to inflation in Turkey is relatively quick and high. (30) On the other hand, recent studies indicate that exchange rate pass-through to both non-tradable and tradable components of CPI inflation in the IT and floating exchange rate environment has declined (Kara et al., 2004, 2007). In this section, we estimate exchange rate pass-through with more up to date data by looking at the dynamic relationships between inflation, exchange rate depreciation, and output gap. Impulse responses obtained from this multivariate VAR framework are then used to calculate exchange rate pass-through. Accordingly the pass-through coefficient is computed as the ratio of cumulative impulse responses of inflation to the cumulative impulse responses of exchange rate depreciation to a shock in the latter. The computation is done for several time intervals to assess the extent and speed of pass-through. (31) As well, we look at the change in the dynamics of exchange rate pass-through to overall CPI inflation and its tradable and non-tradable components during the IT period.

The model is estimated for the period January 1996-June 2007 and includes interactive dummy variables for all variables in the system, which takes the value of zero for the time period prior to the introduction of IIT (January 1996-December 2001) and the value itself of the respective variable during the IIT period (January 2002-June 2007). This would be a convenient method for evaluating changes, if any, in the size and speed of exchange rate pass-through following implementation of the IIT framework. Hence, two sets of impulse responses are considered in calculating pass-through.

Our analysis shows two important findings. Firstly, results indicate that pass-through reaches 34% for the CPI index in a 12-month period with some 86% of that taking place in the first 6 months following the exchange rate shock during January 1996-June 2007 (see Figure 3). Secondly, during the IT period the speed (marginally) and size of pass-through declines. As such, 24% of the exchange rate depreciation is passed on to consumer prices in 12 months with just below 80% of that being completed in 6 months (see Figure 3). (32) Kara et al. (2007) argue that the decline in pass-through is not surprising given the changes to the monetary policy framework and the institutional developments taken place including the operational independence of the CBT.

While lower pass-through is a positive development, Edwards (2006) argues that this 'inflation-centred' view is too simplistic and it tends to ignore the role of relative prices and the real exchange rate. As such, for the exchange rate to act as a shock absorber in general, changes in the nominal exchange rate must be translated into real exchange rate changes. It is thus important to distinguish between two notions of exchange rate pass-through: pass-through into non-tradables and pass-through into tradables. From a policy perspective, pass-through coefficients for tradables and non-tradables should be different, with the former being higher than the latter. It follows then that the adoption of IT regime would improve the shock absorbing capacity of nominal exchange rates if the pass-through from exchange rate to non-tradable prices has declined or if the pass-through to tradable good prices has increased (or, at least, has not declined) as argued by Edwards (2006).

We used the same methodology to estimate the exchange rate passthrough to the non-tradable and tradable parts of the CPI and the impact of the IT regime on the process. (33) This analysis showed two important results. Firstly and as expected, the exchange rate pass-through to tradables is much stronger than that to non-tradables for the entire sample period as well as the IT period. Secondly, there seems to be a significant decline in the speed and extent of pass-through to non-tradable prices while the decline in pass-through to tradable prices seems to be more mild (Figures 4 and 5). While 37.1% and 43.1% of the exchange rate depreciation is passed on to nontradable prices in 6 months and 12 months during the full sample period, the respective figures for the IT period are much lower at 15. (2% and 23.4%. (34) The improvement in the speed of exchange rate pass-through to tradables seems more significant than its magnitude. While three-month cumulative pass-through in the IT period declined to 18% from 30%, the corresponding figures for the 12-month horizon are similar at 42.6% and 46.5%, respectively. While Kara et al. (2007) indicate that the sharp decline in pass-through to non-tradables is a sign of the decline in expectations-driven pass-through, they attribute its more pronounced improvement relative to the tradable CPI component as an indication of the weakening of indexation-driven pass-through.

In summary, there seems to be strong evidence of a change in the dynamics of the exchange rate pass-through that has enhanced the shock-absorbing ability of floating exchange rate regime. The results presented by others and those shown here suggest that the pass-through effect should be less of a concern for policymakers in Turkey and provide a rationale for a relatively weaker reaction of monetary policy to sharp exchange rate movements in the IT period.

'Sudden Stops' and monetary policy reaction

Following the switch to the new monetary framework, the CBT repeatedly emphasised, through regularly published reports as well as through public appearances by officials, that the exchange rate was not a target for monetary policy. Instead, the CBT announced its intention to smooth excessive volatility in the nominal exchange rate without altering the trend movement in the real exchange rate.

Rising confidence in Turkey on the back of disinflation and strong growth, the start of EU accession talks and favourable global conditions in general resulted in a surge in capital inflows during the IIT period (see Figure 10). While foreign investor holdings of Turkish equities and YTL denominated government bonds increased significantly, inflows from foreign direct investment increased sharply as well. Equity and government bond holdings of foreign investor's increased from $3.5 to $42 billion and $3.6 to $21.5 billion, respectively, from the beginning of 2002 to April 2006 (see Figure 11). While total FDI inflows were $13.5 billion during the previous 15 years prior to 2002, the figure from there on to 2007 climbed by another $43 billion. These developments together with de-dollarisation put downward pressure on the nominal exchange rate. While resident's foreign exchange deposits in the banking system stood at over 55% of total bank deposits at end-2002, the figure declined to below 34% by mid-2006.

[FIGURE 10 OMITTED]

[FIGURE 11 OMITTED]

[FIGURE 12 OMITTED]

Given the strength in capital inflows, from 2002 to mid-2007, the CBT's reserve accumulation through discretionary purchases and pre-announced auctions were US$25.2 and US$23.4 billion (net), respectively, as CBT reserves increased from US$19 to over US$65 billion (see Figures 10 and 12). The CBT officials communicated that such discretionary reserve purchases should not be seen as a result of CBT's disapproval of the level of the exchange rate and that buying reserves through auctions mainly reflected the intention to build reserves given the heavy official debt repayment schedule and the desire to improve external risk parameters. Moreover, the timing of initiation, suspension and resumption of auctions was said to be a part of CBT's strategy to curtail excessive volatility through signalling effects. Although the markets did feel on some occasions that the sole purpose of CBT intervention was to prevent the exchange rate from appreciating more than a certain threshold, as pressure in one direction was not perceived as volatility. The CBT partly sterilised the expansion in base money driven by the rise in international reserves (Figure 13). Yet, base money growth remained high and, on average, was three times the rate of annual inflation during the IT period, which partly reflected the rise in demand for real YTL balances.

[FIGURE 13 OMITTED]

While the pressure on the exchange rate was downward in general, the markets were subjected to four sudden stop episodes during the IT period, which resulted in sharp corrections in the exchange rate and the financial markets in general. The first two of these episodes occurred during the April-May period of 2004 and 2005 during which mild portfolio outflows were seen following a period of sharply rising inflows (Figure 14). The 2004 episode saw the exchange rate rise by 17.5%; the sovereign debt yields jump by 190 basis points and benchmark YTL denominated government bond yields increase from 22% to 30%. While local yields and sovereign spreads gradually returned to their pre-correction levels by end-year, the exchange rate saw a volatile period before entering an appreciation trend in early 2005 and returning to its pre-correction level roughly a year later than the start of the episode (Figures 15 and 16). In the meantime, CPI inflation, which had steadily declined to 7.87% from over 70% at the start of 2002, increased to 9.35% by end-2004 but remained below the 12% target.

During the April-May 2005 correction in markets, the exchange rate increased by 9%, sovereign spreads jumped by 235 basis points, while local yields increased marginally from 16% to 18.3% (see Figure 15). Normalisation of yields and spreads took about 3 months while the exchange rate gained back one-third of its loss against the US dollar within 2 months after the initial shock and remained more or less at its new level for the following 12 months (see Figure 15). Inflation continued to decline following the April-May 2005 correction in the markets, as it fell from 8.18% in April 2005 to 7.72% at year-end.

[FIGURE 14 OMITTED]

[FIGURE 15 OMITTED]

Casual observance of the data indicates that the CBT did not react strongly to the first two externally driven corrections in the IIT period (see Figure 16). The sudden stop-driven corrections during April-May in 2004 and 2005 caused the CBT to halt policy rate cuts, as the credibility gap (inflation target less expectations) widened. Yet the CBT refrained from intervention in the foreign exchange market to prevent YTL weakening.

[FIGURE 16 OMITTED]

The mid-2006 episode, the strongest of all sudden stops Turkey was subjected to under the IT framework, resulted in net total portfolio outflows of about 4 billion US dollars, equivalent to about half of cumulative inflows during the prior 12 months of the shock, which led to a 28% correction in the exchange rate, rise in YTL benchmark yields from 13.7% to 21.3% and sovereign spreads by 150 basis points. The shock was mainly driven by negative developments in global liquidity, yet the strength of deterioration in Turkish markets was related to domestic developments as well. The markets seemed to take into account Turkey's rising current deficit (from 0.8% in 2002 to over 7.5% by mid-2006) and its sensitivity to increasing oil prices. But more importantly, April inflation figures released early May came in much higher than market expectations after the CBT had lowered its overnight policy rate on April 27 by 25 basis points. The external shock also coincided with a period of prolonged uncertainty regarding the appointment of the new CBT governor. The government's first choice, which was revealed at the last minute following a period with a caretaker, a vice governor at the time, was vetoed by the president. The entire process was handled very poorly and raised doubts regarding the credibility of the monetary framework shortly after the move to full-fledged IT.

The policy response by the CBT was much stronger this time around, as the overnight policy rate was increased by 400 basis points in a matter of 3 weeks following two emergency MPC meetings on 7 June and 25 June and the CBT sold close to US$2.1 billion in reserves (3.5% of total). Policy rates were increased by another 25 basis points during the 20 July scheduled meeting of the MPC bringing the total tightening to 425 basis points (see Figure 16). By mid-August the exchange rate regained half of the initial weakness before entering a decisive appreciation trend, which took the exchange rate to below its pre-correction level about 12 months later (see Figure 16). Yet, local debt yields remained high, easing marginally and remaining around 19% after 12 months while sovereign spreads declined to pre-correction levels gradually after 12 months (see Figure 15). Inflation increased from 8.16% in May to 9.67%, more than twice the 4% target set for end-2006. There were no discussions of amending the 2006 target following the mid-year external shock and the CBT argued through its official publications that they expected inflation to come close to its 2007 year-end target of 4% following the tightening of monetary policy. Their official forecast saw inflation falling between 9.1% and 10.5% by the end of 2006 with a 70% probability in their third-quarter inflation report issued on September 2006, which did not provide forecasts for end-2007. The first report of 2007 issued in March provided a forecast for end-2007 between 3.6% and 6.6% with a probability of 70%.

The repercussions of a sudden stop of the magnitude, seen in May-June 2006, can be widespread for Turkey. The two immediate effects of the exchange rate correction are the short-term rise in inflation given exchange rate pass-through and slowdown in demand mainly as a result of the affect of depreciation on balance sheets given widespread liability dollarisation. Yet, as discussed in the third section, both where the economy settles in the longer-term and the permanent effect of capital outflows would be heavily influenced by monetary policy reaction. With respect to the latter, for the Turkish case particularly, six issues could have been taken into consideration. Firstly, while still high, the post-IT developments in exchange rate pass-through were encouraging and particularly, the sharp decline in pass-through to the non-tradable component of the targeted CPI index should have provided some comfort to policymakers in terms of the medium-term path in inflation following the external shock. Secondly, the weakening in the exchange rate would likely to generate forces to decrease inflation over the medium-term through balance sheet effects on economic activity. The true extent of the currency mismatches and the balance sheet effects can be difficult to measure. (35) Given the sharp decline in currency mismatches in the banking sector (from $14.5 billion at end-2001 to $0.6 billion at mid2006), the decline in foreign exchange denominated net public sector debt (from $88bn at end-2002 to $30bn at mid-2006), and evidence provided on the foreign currency position of the corporations and households, the aggregate currency mismatches seem to have improved since 2002. We conjecture that aggregate mismatches remained at levels that would unlikely to cause systemic financial distress but that weakening in the currency would decrease economic activity through the same dynamics seen over the past several years. The historical relationship between the real exchange rate and economic growth indicates that appreciation periods and acceleration in economic activity go hand in hand. Likewise, sharp slowdowns in growth are associated with periods of sharp depreciation. Given that the principal constraint to economic growth has been the availability of external financing, it is not surprising to see that the parallel movements in the exchange rate and growth are largely due to fluctuations in capital inflows. Capital inflows lead to exchange rate appreciation, positive balance sheet effects and higher economic growth. Moreover, appreciation has a positive impact on economic activity through lowering production costs of firms where imported inputs are a large part of the production process and related costs constitute a large share of production costs. Likewise, it is through these mechanisms that exchange rate depreciation causes a decline in economic activity, contrary to the conventional wisdom. Indeed, Kara et al. (2007) empirically demonstrate these effects in Turkey. As such, they conclude that the cost channel of the exchange rate on output gap dominates the demand channel. Akyurek and Kutan (2006) also find that real exchange rate changes lead to changes in the output gap in the same direction. Thus, while the sudden-stop-driven exchange rate depreciation would increase short-term inflation mainly through the spike in inflation in the tradable component of CPI, its effects on the output gap would be supportive of disinflation in the medium term. Thirdly, adherence to a floating exchange rate, as emphasised repeatedly by the CBT officials on numerous occasions, would be important in terms of discouraging the expansion of widespread currency mismatches and encouraging wider use of hedging products. Fifthly, the policy response should, to some extent, reflect the awareness of risks associated with carry trade-driven excessive short-term capital inflows and the resulting liquidity expansion. Simply put, large interest rate differentials have been a source of conflict with the IT regime. Finally, given that the 2006 year-end target became simply unattainable in the aftermath of the mid-year external shock, the logical move by the CBT and the government should have been to revise the year-end target. Moreover, given the length of time it takes for the monetary policy transmission mechanisms to effect inflation, especially the aggregate demand channel, a revision of the 2007 inflation target should have been considered as well. Akyurek and Kutan (2006) and Kara et al. (2007) argue that the impact of policy rate changes on economic activity and inflation have become more predictable and in the direction in line with theory, improving the transmission capacity of monetary policy following the change in the monetary regime to IT. Upward revision of the 2007 target would have amounted to softening the targeted disinflation trajectory since the year-end target for 2008 was set at 4% as well. The CPI inflation climbed from 8.83% in April 2006 to over 11% within 4 months before declining to the 7.3% level as of August 2006. The return of inflation to pre-shock levels was a welcome development. And yet, despite the significant tightening of monetary policy, inflation was still significantly above the path leading to the 4% target. As well and as expected, the deceleration in growth became visible only in the second quarter of 2007, that is during the third quarter following the external shock. These developments are supportive of the argument for the softening of the targeted disinflation trajectory following the external shock of mid-2006.

These complications highlight the view that monetary policy in the real world is more of an art than a science and there is very little doubt that engineering an optimal monetary policy response was a challenging task under the circumstances following the May-June 2006 sudden stop episode. Yet, increasing interest rates at the speed and extent they were is a policy response that should be analysed carefully to the extent that the move seemed to aim at limiting the short-term temporary spike in inflation and containing negative balance sheet effects through supporting a recovery in the exchange rate. The issue addressed here is one of magnitude. Was the policy response associated with adherence to an open economy IT regime or due to fear of floating type of reaction which imply a hidden policy goal of managing the exchange rate? (36) While the extent and scope of monetary policy reaction in mid-2006 may suggest a defense of the currency during this episode, particularly when compared to the other three episodes experienced during the sample period, we have not attempted to test statistically as to whether the CBT has reacted to exchange rate corrections above and beyond their effects on inflation. To test this possibility the dynamic evolution of the policy function must be analysed and at this stage given the shortness of the data period and the ongoing structural transformation in the economy, meaningful statistical analysis is difficult to conduct.

CONCLUSION

Turkey's adoption of an IT-like regime in 2002 occurred when inflation was on the rise and after banking sector recapitalisation process had led to a significant jump in public sector debt. Concerns over public sector debt rollover were dominating the dynamics of financial markets with strong repercussions on macroeconomic fundamentals. Fiscal discipline, the improvement in the policy environment and the overall progress in the economy provided progressively stronger support to the enhancement of the new monetary regime, which lent support to the CBT's decision of a gradual shift to full-fledged IT after 3 years. The change provided further transparency to the monetary policy framework and added more flexibility to monetary policy.

Evidence from the estimation of a Taylor-like central bank reaction function shows that the CBT lowered real rates mainly in conjunction with the decline in Turkey's risk premium, as inflation declined significantly during 2002-2006. While changes in the exchange rate, output gap and inflation expectations were found to influence policy rate changes, during this period of structural transformation and transitional dynamics, it seems that the more comfortable the CBT felt with progress on fiscal policy and reforms, the easier and quicker seemed to be the move to lower real rates.

With the introduction of a comprehensive economic programme, implementation of structural reforms, completion of large privatisation projects and the start of accession talks with the EU, Turkey attracted significant amount of foreign capital. As a result of this, during the inflation targeting and floating exchange rate period, the YTL remained strong and the real exchange rate appreciated significantly. The latter together with high growth led to a significant widening of the current account deficit, which rendered Turkish financial markets susceptible to external shocks. The markets were subjected to externally driven reversal of capital inflows on three occasions. While monetary responses to the shocks in April-May of 2004 and 2005 were relatively limited, a much stronger reaction was given to the greater turbulence of mid-2006. The policy reaction of mid-2006 brings the question of whether or not this response was associated with adherence to an open economy IT regime or due to fear of floating type of reaction geared towards bringing a recovery in the exchange rate. Given the extent of tightening in the policy rate, this question is particularly relevant in the Turkish case where the exchange rate depreciations have been associated with economic slowdowns.

Inflation returned to its pre-shock level within a year, yet after having overshot the 2006 target, it remained significantly above the 2007 target as well despite the sharp rise in policy rates. The developments in the inflation rate are supportive of the argument for the softening of the targeted disinflation trajectory following the external shock of mid-2006.

REFERENCES

Akyurek, C. 1999: Post-liberalization inflation in Turkey: an empirical investigation. Yapi Kredi Economic Review 10(2): 31-53.

Akyurek, C. 2006: The Turkish crisis of 2001: A classic? Emerging Markets Finance and Trade 42(1): 5-32.

Akyurek, C and Kutan, A. 2006: Can inflation targeting be effective in emerging economies?: The Turkish experience. Working Paper No. 06-0902, Southern Illinois University Edwardsville, Economics and Finance Department.

Amato, JD and Gerlach, S. 2002: Inflation targeting in emerging market and transition economies: Lessons after a decade. European Economic Review 46(4-5): 781-790.

Armenio, F, Goldfajn, I and Minella, A. 2003: Inflation targeting in emerging economies. NBER Working paper series No. 10019, NBER: Cambridge, MA, October.

Ball, CP and Reyes, J. 2004: Inflation targeting or fear of floating in disguise: The case of Mexico. International Journal of Finance and Economics 9: 49-69.

Ball, CP and Reyes, J. 2006: Inflation targeting or fear of floating in disguise: A broader perspective. Journal of Macroeconomics, forthcoming.

Ball, L. 1999: Policy rules for open economies. In: Taylor, J (ed). Monetary Policy Rules. University of Chicago Press: Chicago pp. 43-54.

Blejer, M, Leone, AM, Rabanal, P and Schwartz, G. 2001: Inflation targeting in the context of IMF-supported adjustment programs. IMF Working Paper, 01/31.

Borensztein, E and De Gregorio, J. 1999: Devaluation and inflation after currency crises. University of Chile, unpublished manuscript. Available at http://www.itam.mx/lames/papas/invitses/degregor.pdf.

Brash, DT. 2002: Inflation targeting: New Zealand's experience over 14 years (2002). The North American Journal of Economics and Finance 13(2): 99-112.

Brimmer, AF. 2002: Central banks and inflation targeting in perspective. The North American Journal of Economics and Finance 13(2): 93-97.

Cabellero, RJ and Krishnamurthy, A. 2003: Inflation targeting and sudden stops. NBER Working Papers 9599.

Calvo, GA and Reinhart, C. 2002: Fear of floating. Quarterly Journal of Economics 117(2): 379-408.

Campa, JM and Goldberg, LS. 2002: Exchange rate pass-through into import prices: a Macro or Micro phenomenon? NBER Working Paper Series No. 8934.

Carare, A, Schaechter, A, Stone, MR and Zelmer, M. 2002: Establishing initial conditions in support of inflation targeting. IMF Working Paper.

Carare, A and Stone, MR. 2006: Inflation targeting regimes. European Economic Review 50(5): 1297-1315.

Carranza, L, Sanchez, JEG and Biscarri, JG. 2004: Exchange rate and inflation dynamics in dollarized economies. Department of Economics, Universidad de Navarra Working Paper No. 10/04.

Cespedes, LF, Chang, R and Velasco, A. 2000: Balance sheets and exchange rate policy. NBER Working Paper Series No. 7840.

Choudhri, E and Hakura, D. 2006: Exchange rate passthrough to domestic prices: Does the inflationary environment matter? Journal of International Money and Finance 25(4): 614-639.

Dodge, D. 2002: Inflation targeting in Canada: Experience and lessons. The North American Journal of Economics and Finance 13(2): 113-124.

Edwards, S. 2006: The relationship between exchange rates and inflation targeting revisited. Central Bank of Chile Working Papers No. 409.

Eichengreen, B. 2002: Can emerging markets float? Should they inflation target?. University of California: Berkeley.

Goldfajn, I and Werleng, S. 2000: The pass-through from depreciation to inflation: A panel study. Working Paper 423, Potifical Catholic University of Rio de Janeiro.

Goldstein, M and Turner, P. 2004: Controlling currency mismatches in emerging markets. Institute for International Economics: Washington, DC April 2004.

Golineli, R and Rovelli, R. 2005: Monetary policy transmission, interest rate rules and inflation targeting in three transition countries. Journal of Banking and Finance 29(1): 183-201.

Gomez, J. 2004: Inflation targeting, sudden stops, and the cost of fear of floating. The Central Bank of Columbia.

Goncalves, CES and Salles, JM. 2008: Inflation targeting in emerging economies: What do the data say? Journal of Development Economics 85(1-2): 312-318.

Honda, Y. 2000: Some tests on the effects of inflation targeting in New Zealand, Canada, and the UK. Economics Letters 66(1): 1-6.

Johnson, DR. 2002: The effect of inflation targeting on the behavior of expected inflation: Evidence from an 11 country panel. Journal of Monetary Economics 49(8): 1521-1538.

Jonas, J and Mishkin, FS. 2003: Inflation targeting in transition countries: Experience and prospects. NBER Working Paper 9667.

Kara, H, Kucuk-Tuger, H, Ozlale, U, Tuger, B and Yucel, EM. 2007: Exchange rate regimes and pass-through: Evidence from the Turkish economy. Contemporary Economic Policy 25(2): 206-225.

Kara, H and Ogunc, F. 2004: Exchange rate pass-through in Turkey: It is slow but is it really low? Central Bank of Turkey Working Paper 05/10.

Kara, H., Ogunc, F., Kucuk-Tuger, H., Ozlale, U. and Sarikaya, C.. 2007: Estimating the output gap in a changing economy. Central Bank of Turkey Working Paper.

Leigh, D and Rossi, M. 2002: Exchange rate pass-through in Turkey. IMF Working Paper No. 204.

McCarthy, J. 1999: Passthrough of exchange rates and import prices to domestic inflation in some industrialized economies. BIS Working Paper No. 79.

Mishkin, FS. 2000: Inflation targeting in emerging market countries. NBER Working Paper Series, http://www.nber.org/papers/w7618.

Mishkin, FS. 2004: Can inflation targeting work in emerging market countries. NBER Working Paper No. 10646.

Nogueira Jr, RP. 2006: Inflation targeting, exchange rate passthrough and fear of floating Department of Economics, University of Kent Studies in Economics No.0605.

Orlowski, LT. 2004: Money rules for monetary convergence to the euro. Journal of Policy Modeling 26(7): 817-837.

Orlowski, LT. 2005: Monetary convergence of EU accession countries to the eurozone: A theoretical framework and policy implications. Journal of Banking and Finance 29(1): 203-225.

Orlowski, LT and Rybinski, K. 2006: Implications of ERM2 for Poland's monetary policy. Economic Systems 30(4): 346-365.

Saatcioglu, C and Korap, L. 2006: Determinants of Turkish inflation. Discussion Paper No. 7, Turkish Economic Association.

Schaechter, A, Stone, MP and Zelmer, M. 2000: Adopting inflation targeting: practical issues for emerging market countries. Occasional Paper No. 202, IMF.

Schmidt-Hebbel, K and Tapia, M. 2002: Inflation targeting in chile. The North American Journal of Economics and Finance 13(2): 125-146.

Schmidt-Hebbel, K and Werner, AM. 2002: Inflation targeting in Brazil, Chile, and Mexico: Performance, credibility, and the exchange rate. Central Bank of Chile Working Papers.

Siklos, PL and Abel, I. 2002: Is Hungary ready for inflation targeting? Economic Systems 26(4): 309-333.

Taylor, J. 1993: Discretion versus policy rules in practice. Carnegie Rochester Conference Series on Public Policy 39: 195-214.

Tortes, A. 2003: Monetary policy and interest rates: Evidence from Mexico. The North American Journal of Economics and Finance 14(3): 357-379.

(1) Recent studies include Brimmer (2002), Dodge (2002), Johnson (2002) and Carare and Stone (2006).

(2) For individual country experiences, see Brash (2002) and Honda (2000) for the New Zealand; Armenio et al. (2003) for Brazil; and Torres (2003) for Mexico. Ainato and Gerlach (2002) provide evidence from many developing countries and emerging markets. More recently, Goncalves and Salles (2008) provide evidence from 36 emerging markets and they find that those countries that followed an IT regime did have a better economic performance than those that did not.

(3) The following studies study the experience of the new member states: Amato and Gerlach (2002), Siklos and Abel (2002), Jonas and Mishkin (2003), Golineli and Rovelli (2005), Orlowski (2004, 2005) and Orlowski and Rybinski (2006).

(4) See Schaechter et al. (2000), Jonas and Mishkin (2003) and Carare et al. (2002) for a discussion of requirements for a successful IT regime.

(5) For an analysis of the role of exchange rate movements and in particular of external shocks on inflation dynamics, see Akyurek (1999).

(6) The gross debt ratio increased from 63% at end-2000 to 101% by end-2001, as the share of floating rate notes and foreign exchange denominated and linked debt in total domestic bonds reached 75%.

(7) During the first half of 2002, the IMF programme was on hold and at the same time the US had offered Turkey a financial package that could have provided in excess of US$30 billion in return for passing a parliamentary motion allowing US troops' entry to Iraq through Turkey. The rejection of the motion came as a big surprise and caused extreme volatility in the Turkish markets.

(8) Sharp increases in the primary surplus were achieved by increases in indirect taxation and cutting investments in the past during difficult economic times. Yet, fiscal discipline unraveled shortly after conditions stabilised, as challenging reforms regarding taxes and social security were never completed, which were also a part of the ongoing IMF programme at the time IIT was introduced. It was important for the CBT to be able to disentangle temporary from permanent influences on the budget balance in order to gauge the medium-term orientation of fiscal policy.

(9) The costs of recapitalising public sector and private sector banks were approximately 15% of GDP for each, as the total reached 35%.

(10) Major improvements included the incorporation of market risks in capital adequacy calculations, imposing strict limits on newly defined related-party lendings, setting of new rules and supervisiory practices to limit/control foreign exchange exposure, implementation of new provisioning rules and limits on non-financial participations.

(11) Short foreign exchange positions of the banking sector declined significantly from US$15 billion in 2000 to US$0.6 billion in 2002.

(12) Major takeovers/mergers involved big European names such as Unicredito of Italy (50% of Koc Financial Services), Fortis of Belgium (100% of Disbank), BNP paribas of France (50% of Turkiye Ekonomi Bankasi) and Rabobank of Netherlands (36.5% of Seker Bank) as well as USA-based GE Consumer finance (26% of Garanti Bank).

(13) Inflation targets for year-end 2006, 2007 and 2008 were set at 5%, 4% and 4%, respectively, with the year-end target for 2009 also kept at 4%.

(14) In this type of set up, the main role of the NIR floor is to indicate whether the programme is likely to achieve its external objective, while the ceiling on NDA seeks to ensure that this objective is not jeopardised by excessive credit expansion and sterilised intervention (Blejer et al., 2001).

(15) This change seems to have led to the need to adapt different monetary conditionality to take into account the specific features of the full-fledged IT framework. Accordingly, an inner and an outer tolerance band is established around the central target path for conditionality purposes. The former is 1% and if it is breached (at quarter-end or year-end), this will require an informal consultation with the IMF on policy response. The outer band is 2% around the central path and if it is breached there will be a formal review of policies by the IMF Board.

(16) The IT regime has an advantage such that a stable relationship between money and inflation is not critical to its success. The strategy does not depend on such a relationship, but instead uses all available information to determine the best settings for the instruments of monetary policy.

(17) Note that [phi] = 1/([alpha][beta]) and [chi] = (1 + [lambda])/[beta].

(18) A value of zero to [chi] would correspond to a strict IT regime.

(19) The starting point of the derivation of open economy equations is the interest parity condition: [e.sub.t] = E([e.sub.t+1]) = [r.sub.t] - r' + [V.sub.t], where E is an expectations operator and r' and [V.sub.t] are foreign interest rate and pure portfolio disturbances (Eichengreen, 2002).

(20) Likewise the adjustment in the policy rate to a given output gap would be smaller, the more sensitive aggregate demand is to interest rate changes and the weaker is the level of persistence in output.

(21) Eichengreen (2002) also argues that liability dollarisation may lead to an extreme case where balance sheet effects are so large that [delta] turns (-) and [delta] > [beta]. When the exchange rate depreciates by a large amount the balance sheet effects may dominate, but when it depreciates by a small amount the favorable competitiveness effects may dominate. In other words, for a range of depreciations, the weakening in the domestic currency may have an expansionary affect on output whereas depreciations above a threshold may lead to contraction in output. The latter case occurs when the sharp depreciation does little to enhance competitiveness because of the speed with which it is passed through to inflation. So if the exchange rate weakens sufficiently the CB will react strongly to bring back the currency and avoid severe balance sheet effects on banks and non-banks. If the depreciation is mild then the response will be mild as well where the CB will allow the exchange rate to adjust to a new long-run equilibrium.

(22) Cabellero and Krishnamurthy (2003) argue that the central bank could target inflation in non-tradables. The advantage of this is that it prevents the central bank from giving excessive attention to the exchange rate on inflation in the short-term, and it enables the central bank to pay due attention to inflation in the medium term and countercyclical monetary policy. Another policy option is to introduce escape clauses to the IT framework allowing for the target to be amended in the face of shocks to the exchange rate.

(23) See Borensztein and De Gregorio (1999), Goldfajn and Werleng (2000), Schmidt-Hebbel and Tapia (2002).

(24) See for example Nogueira (2006) and Choudhri and Hakura (2006).

(25) They model the sudden stop as a tightening of international financial constraints.

(26) There were significant delays in completion of IMF programme reviews during 2002-2006.

(27) Real rates are obtained by deflating nominal overnight policy rates by 12-month forward-looking inflation expectations based on CBT's survey among private sector economists. Monthly output gap values were derived from quarterly figures provided by the CBT using a quadratic matching of the average.

(28) Although we thank an anonymous referee for pointing out that the fact that the risk premium could change in response to inflation, output and exchange rate developments, these factors may have still affected policy rate decisions.

(29) See Saatcioglu and Korap (2006) for a comprehensive review of the literature on Turkish inflation.

(30) See for example Leigh and Rossi (2002).

(31) For example pass-through after 12 months corresponds to the ratio of the cumulative impulse responses of inflation to cumulative impulse responses of depreciation during the 12 periods following the exchange rate shock. This methodology, based on McCarthy (1999), has been widely used and was also the approach adopted by the CBT research department.

(32) Our results are similar to those of the CBT (see Kara and Ogunc, 2004).

(33) Since the State Institute of Statistics does not report a tradable and non-tradable breakdown of the CPI index, we used our own estimates.

(34) Our findings are similar to those of Kara et al. (2007) in terms of the change in the dynamics of pass-through for the overall CPI index as well as for its tradable and non-tradable components.

(35) The proper measurement, the aggregate effective currency mismatch, as defined by and presented in Goldstein and Turner (2004), shows that mismatches have peaked in Turkey in 2002 (page 50).

(36) Schmidt-Hebbel and Werner (2002) show that countries may exhibit a temporary fear of floating during particular periods or events.

CEM AKYUREK (1) & ALI M KUTAN (2)

(1) Istanbul, Turkey. E-mail: cemakyurekl@gmail.com

(2) Department of Economics and Finance, Southern Illinois University Edwardsville, Edwardsville, IL 62026-1102, USA. E-mail: akutan@siue.edu

Table 1: Estimates of the Central Bank reaction function--equation 6

Dependent variable: REALRATE
Method: Least squares
Sample (adjusted): 2003:04, 2007:06
Included observations: 51 after adjustments

Variable Coefficient Std. error t-Statistic

C 0.03886 0.01187 3.2754
Inflation Gap 0.42715 0.26236 1.6281
Exchange rate Depreciation 0.04403 0.02214 1.9884
Output Gap 0.24680 0.09180 2.6884
Risk Premium (EMBITR) 2.29398 0.38783 5.9150
AR(1) 1.06049 0.15322 6.9215
AR(2) -0.27058 0.14433 -1.8747

Adjusted [R.sup.2] 0.964445 F-statistic 227.04

Variable Prob.

C 0.002
Inflation Gap 0.111
Exchange rate Depreciation 0.053
Output Gap 0.010
Risk Premium (EMBITR) 0.000
AR(1) 0.000
AR(2) 0.068

Adjusted [R.sup.2] 0.000