WPS4691 Policy ReseaRch WoRking PaPeR 4691 Comparing Constraints to Economic Stabilization in Macedonia and Slovakia: Macro Estimates with Micro Narratives Martin Melecky Evgenij Najdov The World Bank Europe and Central Asia Region Poverty Reduction and Economic Management Department August 2008 Policy ReseaRch WoRking PaPeR 4691 Abstract This paper re-emphasizes the link from structural Macedonia to those faced by Slovakia, the authors policies to enhanced macroeconomic stabilization using find that Macedonia has relatively weaker transmission a small structural model estimated on quarterly data for mechanisms of monetary policy, higher output rigidity, Macedonia and Slovakia over 1995-2007. The success and a lower exchange rate pass-through, and faces of macroeconomic stabilization, typically in the hands larger external shocks. For Macedonia, these relatively of monetary policy, is not only determined by a suitable higher constraints on monetary policy together with the choice of the nominal anchor, which shapes the reaction chosen exchange rate anchor result in greater output and function of monetary policy, but also the constraints inflation volatility relative to Slovakia. Hence, it appears within which the monetary policy strives to achieve that small, open economies with stronger economic its objectives. The key attributes of the constraints to rigidities should apply monetary policy regimes that allow macroeconomic stabilization are economic rigidities for more flexible adjustments in external relative prices to and structural shocks. By benchmarking the estimated enhance their macroeconomic stability. economic rigidities and structural shocks faced by This paper--a product of the Poverty Reduction and Economic Management Department, Europe and Central Asia Region--is part of a larger effort in the department to understand the tradeoffs and complementarities between pro-growth structural and stabilization policies. Policy Research Working Papers are also posted on the Web at http://econ.worldbank. org. The author may be contacted at mmelecky@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Comparing Constraints to Economic Stabilization in Macedonia and Slovakia: Macro Estimates with Micro Narratives* Martin Melecky Evgenij Najdov World Bank World Bank Key words: Transition economies, Macedonia, Slovak Republic, macroeconomic stabilization, economic rigidities, New Keynesian Policy model JEL Classification: E30, E58, D11, D21, D53 * We thank Eduardo Ley, Ivailo Izvorski, Ardo Hansson and Bernard Funck for their comments and suggestions on the earlier draft of the paper. All remaining errors are ours. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Contacts: mmelecky@worldbank.org; enajdov@worldbank.org. 1 1. Introduction The link from macroeconomic stability to economic growth has been studied for some time in both academic and policy circles, most recently e.g. by Loayza et al. (2005) and Iradian (2007). The reverse link from better structural, pro-growth policies to enhanced macroeconomic stabilization is emphasized less often but is equally obvious. Stabilization policy is focused on maintaining price stability and unemployment levels close to the natural unemployment rate. These goals are typically outsourced by government to, preferably, an independent agent, the central bank which chooses a (explicit or implicit) nominal anchor, as an intermediate goal and a way to anchor the public's expectations. The chosen nominal anchor then dominates in a reaction function of the central bank, i.e. the central bank adjusts its monetary policy instrument by putting the highest weight on sticking to the nominal anchor. The typical representations of the central bank's reaction function, which we also consider in this paper, are due to due to Taylor (1993) for (explicit or implicit) inflation targeting, and Benigno et al. (2007) for exchange rate targeting. Both representations of the reaction function happen to be interest rate rules with a certain degree of discretion. Which of the two reaction functions will be more successful in stabilizing a given economy depends on the constraints that the economy places on achieving the monetary policy objectives. These constraints, in turn, can be significantly alleviated by appropriate structural reforms, which mainly promote the economy's flexibility and responsiveness to changes in the monetary policy stance. In this paper we consider two transition economies ­ Macedonia and the Slovak Republic ­ which happen to apply exchange rate targeting and inflation targeting regimes, respectively. We assume that the economies' constraints to achieving the monetary policy objective of low inflation and output variability are represented by the consumption behavior of households, pricing behavior of firms, and interactions of the domestic economy with the external sector. We describe the above constraints using a small open-economy model with rational expectations where the model parameters are estimated using quarterly data for Macedonia and the Slovak Republic over 1995-2007. In the assumed model, the key elements of the constraints on monetary policy are real and nominal rigidities, and domestic and external structural shocks. We benchmark the differences in the estimated model coefficients and provide some explanations for the 2 estimated differences using relevant microeconomic data. The gaps between Macedonian and Slovak structural indicators show that the economic rigidities in Macedonia could be significantly diminished by appropriate structural reforms. While structural shocks by their nature cannot be generally reduced through government policies, reducing fiscal shocks, i.e. the magnitude and frequency of changes in the fiscal policy stance, is in control of government policy. More specifically, we find that for Macedonia vis-à-vis the Slovak Republic, the consumption habit formation and cost of adjusting capital stock are higher, the credit channel of monetary policy is weaker, the net exports elasticity to exchange rate changes is weaker, the portion of non-Ricardian households is higher, the effect of increasing capacity utilization on inflation is higher, and the exchange rate pass- through is lower. On the other hand, we find that Macedonia does better in terms of containing fiscal policy shocks than the Slovak Republic. Overall, it appears that the Slovak Republic is enjoying a lower inflation/output volatility tradeoff compared with Macedonia, for which our gap analysis provides some directions for improvements in the structural reforms. In addition, to our knowledge, this paper makes a first attempt to fit a structural model with rational expectations to the data on the Macedonian and Slovak economies. The remainder of the paper is organized as follows. Section two discusses the main features of the employed structural model. Section three describes the data and the estimation methodology. Section four discusses the estimation results. Section five offers microeconomic explanations of the estimated macroeconomic characteristics of the Macedonian and Slovak economies. Section six analyzes impulse responses of selected economic variables to identified structural shocks. And, section seven concludes. 2. Model Description This section describes theoretical underpinnings of the New Keynesian policy model that we estimate to capture some fundamental characteristics of the Macedonian and Slovak economies. Let Et Xt +1denote the rational expectation forecast of Xt +1 conditional on the 3 information set available to the forecasting agent at time t. The equation describing inflation dynamics is modeled by the following "hybrid" Phillips curve1 t = Ett +(1- )t +1yt +2st +AS +1 -1 (1) ,t where t is CPI inflation, yt is the output gap, st is the change in the nominal effective exchange rates (an increase implies depreciation of the Macedonian denar ­ MKD; or the Slovakian koruna ­ SKK), and AS is an autocorrelated aggregate supply (AS) shock. ,t Although allowing for an inertial effect by giving a non-zero weight to t -1in equation (1) was initially empirically motivated, the effect can be derived from a staggered price- setting mechanism, where a proportion of firms use a naïve, backward-looking rule to forecast inflation. The inertial effect also arises as a consequence of a Calvo-type price setting mechanism, with partial indexation to last period's inflation. For explicit derivation of the hybrid Phillips curve, see e.g. Christiano et al. (2005). The empirical usefulness of the hybrid specification has been advocated in e.g. Fuhrer and Moore (1995). Further, CPI inflation increases in response to a positive output gap and thus increasing marginal cost of production. The effect of the exchange rate on CPI inflation is exercised directly through the domestic currency price of imported final goods, and the domestic currency price of the imported intermediate inputs. Eventually, the exchange rate will also affect nominal wages via the effect of CPI inflation on wage setting. In either case, the exchange rate will affect the cost of domestically produced goods and inflation in the prices of domestically produced goods (see e.g. Svensson, 2000). For empirical reasons we use the specification with the first difference in the real exchange rate as in Giordani (2004). The output gap dynamics is described by the following aggregate demand (IS) equation: yt = yEt yt + 1- y yt +1 it - Et t +2qt +3Etgt + IS +1 ( ) -1 ( - j - j ) -k +l (2) ,t 1The term hybrid relates to the fact that the Phillips curve is backwards, as well as forward-looking in inflation. 4 where it is the nominal interest rate, qt is the change in the real effective exchange rate, gt is the change in government consumption per GDP, and IS an autocorrelated ,t aggregate demand shock. IS is assumed to encompass both domestic and foreign ,t demand shocks. One can see from equation (2) that the output gap depends on its expected value one period ahead and its lagged value, where the relative impact is determined by the size of y . The forward-looking term is due to households' inter- temporal optimizing behavior and the lagged term arises as a result of consumption-habit formation, or a costly adjustment of the capital stock under inter-temporal optimization, see e.g. Clarida et al. (2002) and Christiano et al. (2005) for further details. When the interest rate increases, consumption today in terms of consumption tomorrow becomes more costly, leading to a reduction in current domestic demand. Moreover, the interest rate affects the user cost of capital, influencing investment demand. Aggregate demand is thus influenced through intertemporal substitution effects (by the real interest rate), and through intratemporal price effects (by changes in the real exchange rate). The presence of the real exchange rate captures the resulting changes in export (and import) demand. More specifically, the net exports are assumed to increase with depreciating domestic currency (increase in qt ). Further, increased government consumption per GDP, gt , can have either positive or no effect on the output gap depending on the proportion of the rule-of-thumb consumers in the economy. A higher portion of the rule-of-thumb consumers, i.e. non-Ricardian households will result in higher effect of changes in government consumption on the output gap, ceteris paribus (Gali et al., 2007).2 The motivation for the open-economy IS equation can be found in Monacelli (2005), Clarida et al. (2001), and Svensson (2000). The lag length selection of variables, i.e. determination of j, k and l in the considered ranges of j = 0... - 2 , k = +1... -1 and l = +1... -1, respectively, is motivated empirically, and has been performed in an encompassing manner using the Bayesian Information Criterion (BIC). In the lag-length selection process, we have imposed the restriction that the impact of exchange rate changes on the output gap is faster than the impact of real interest rate changes. The latter 2If government investment was included instead, one could explore the possibility of either crowd in or crowd out effects on private domestic demand. We however choose to stick to the specification of fiscal policy analogous to Gali et al. 5 is a stylized fact which holds for most small open economies (see e.g. Buncic and Melecky) and we assume it for Macedonia and the Slovak Republic as well. When choosing the suitable monetary policy reaction function for Macedonia and the Slovak Republic we turn to the classification of monetary policy regimes by the IMF3. The IMF classifies the monetary policy regime of Macedonia as a fixed exchange rate one (with respect to the EUR). This regime has been applied by Macedonia since 1995 with a single devaluation in 1997. The chosen reaction function is thus of the following form it = it + st + MP * (3) ,t advocated by Benigno et al. (2007). The monetary policy rule in equation (3) postulates that the National Bank of the Republic of Macedonia (NBRM) adjust its interest rate in response to changes in the foreign interest rate, it , and the nominal exchange rate while * applying some degree of discretion to this rule, as represented by MP . The latter is,t assumed to be a white-noise process.4 On the other hand, the monetary policy regime applied by the Slovak National Bank (SNB) has been classified as inflation targeting since December 2004, and recently applied within the context of ERMII which the Slovak Republic joined in 2007. Nevertheless, implicit inflation targeting was applied by the SNB since 1998 when the pegged exchange rate framework was abandoned and a combination of managed floating and implicit inflation targeting was adopted. Inflation targeting is traditionally represented by the Taylor rule in the kind of models used in this paper. The Taylor rule has been found empirically plausible and reasonably robust to different model structures (see Svensson, 2000). In some circumstances, the Taylor rule can also be used to describe optimizing behavior (see Benigno and Benigno, 2003). A forward-looking version of the Taylor rule is employed to emphasize a central bank's focus on future inflation when adjusting its monetary policy instrument. We also allow for explicit reaction to changes in the exchange rate in the view of possible exchange rate smoothing, and most recently 3http://www.imf.org/external/np/mfd/er/2006/eng/0706.htm#table. 4An i.i.d specification of the monetary policy shock is a common assumption in the literature, see Smets and Wouters (2003) and Del Negro et al. (2005). 6 the ERMII rules5, and the existence of fixed exchange rate regime in the pre-1998 period. Since the output gap appeared to be empirically insignificant we ended up using the following modified Taylor rule it = iit + (1- i )(Ett + st )+ MP . -1 +1 (4) ,t The specification in equation (4) implies that the monetary authority responds to expected inflation one period ahead and the current changes in the nominal exchange rate, while at the same time adhering to a certain degree of inertia in it . The dynamics of fiscal policy stance, in terms of changes in government consumption per GDP, is described simply by a first-order autoregressive process (AR(1)) gt = g gt + G -1 (5) ,t where G is the idiosyncratic change in government consumption per GDP which is ,t allowed to be autocorrelated. Finally, the evolution of the real effective exchange rate is specified in order to close the model. The change in the real exchange rate, in logs, is defined as qt st -t +t where t is unobserved foreign inflation. We adopt an assumption * * common in the literature of the exchange rate evolving according to real UIP. The UIP condition is generally stated as an identity over the log of the exchange rate and interest rates, with the exchange rate expressed as the ratio of domestic to foreign currency units. Etqt = (it - Ett+1)-rt * (6) where rt is the unobserved foreign real interest rate. rt thus comprises wider range of * * external shocks including the forward exchange rate risk premium and the terms of shock, and is possibly serially correlated.6 5These include stabilization of the LC/EUR exchange rate around chosen exchange rate parity within 15% bands. 6For more details regarding the empirical properties of UIP, see the studies by Ferreira and Leon-Ledesma (2007), Chinn and Meredith (2004), and Mark and Moh (2001). 7 3. Data and the Estimation Method In order to maximize the available data coverage while considering the data quality, we use quarterly data for Macedonia from 1997Q1 to 2007Q3, and for the Slovak Republic from 1995Q1 to 2007Q3. All data for the Slovak Republic were obtained from the IMF's International Financial Statistic except the nominal interest rate which was obtained from Datastream. For both countries, the output gap was constructed as a deviation of quarterly real GDP in logs from its potential levels estimated using the Hodrick-Prescott filter. The GDP series for Macedonia was obtained from the National Statistical Office. Inflation was calculated as an annualized percentage change in quarterly CPI, which for Macedonia was obtained from the National Statistical Office. The interest rate used for the Slovak Republic is the three-month interbank rate (middle rate) from Datastream. This rate tracks well the current monetary policy rate (the repo rate), on which data are available only since 2000. The interest rate used for Macedonia is the Central Bank Bill rate obtained from the National Bank of Macedonia. We have not used the interbank rate for Macedonia as the interbank money market was quite inactive over the analyzed period. The observable exchange rate employed is the real effective exchange rate that for Macedonia was obtained from the National Bank of Macedonia. The series of government consumption for Macedonia is readily available on the quarterly basis only from 1999 onwards, and we have extrapolated the series back to 1997Q1 using a constructed series of government spending on wages and salaries, and goods and services. The correlation coefficients between the levels and differences of the constructed series and the actual government consumption series over 1999Q1-2007Q4 is 0.95 and 0.98, respectively. Both government consumption series and the spending of government on wages and salaries, and goods and services were obtained from the Macedonian National Statistical Office. In addition, when calculating the ratio of government consumption to GDP we have used the CPI index to convert the real GDP series into current prices because nominal GDP is not available for Macedonia on a quarterly basis. All data are demeaned prior to the estimation. Giordani (2004) has recently pointed out that working with demeaned data avoids dealing with parameter instability and structural breaks which, he finds, largely affect the unconditional mean of the modeled series. 8 There are several estimation methods used in the literature to fit New Keynesian models to data. The recently most popular method has been the Bayesian estimation which overcomes some problems of the Full Information Maximum Likelihood method7 by imposing priors on the structural coefficients' distributions (see e.g. An and Schorfheide, 2005; or Buncic and Melecky, 2008). Given the relatively higher model uncertainly for transition economies, and the fact that system estimators can be inconsistent if one of the equations in the system is misspecified (see Johansen, 2005) our preferred estimation method is the Generalized Method of Moments (GMM) (see e.g. Gali and Gertler, 1999; among others). We used two lags of the variables in the system as instruments. The long-run heteroscedasticity and autocorrelation consistent (HAC) covariance matrix weighting the moment conditions in the GMM estimator is estimated using the Quadratic kernel with the variable New-West bandwidth selection. In addition, pre-whitening of the moment conditions was applied. 4. Discussion of Estimation Results The estimated model parameters, using GMM and the quarterly data for Macedonia from 1997Q1-2007Q3 and for the Slovak Republic from 1995Q1-2007Q3, are reported in Table 1.8 7 One of the drawbacks of using ML is that parameters can take on corner solutions or theoretically implausible values. Additionally, it is often the case that the log-likelihood function is flat in certain directions of the parameter space and extremely hilly overall, so that without careful constraints on the parameters space, it is difficult to numerically maximize the log-likelihood function (see the discussion in An and Schorfheide, 2005, for more details). 8An explicit, microfounded derivation of the kind of model presented here could be found in e.g. Svensson (2000) or Monacelli (2005). Examples of the model's estimations using data for Australia, Japan, the U.S., or the euro area could be found in e.g. Melecky (2008) or Buncic and Melecky (2008). 9 Table 1: Estimates of the model parameters using GMM parameter Estimate for Macedonia Estimate for Slovakia y 0.4764 (0.0354)*** 0.6504 (0.0871)*** 1 0.0130 (0.0007)*** j = -1 0.0375 (0.0078)*** j = -2 2 0.0623 (0.0226)*** 0.1229 (0.0389)*** 3 0.2343 (0.0611)*** 0.1408 (0.0340)*** 0.5657 (0.0069)*** 0.5660 (0.0254)*** 1 0.0394 (0.0015)*** 0.0145 (0.0070)** 2 0.0240 (0.0032)*** 0.0374 (0.0159)** i na 0.6427 (0.0273)*** na 1.0152 (0.0698)*** 3.7619 (1.2333)*** 0.3962 (0.0503)*** g -0.0943 (0.1573) -0.3317 (0.1252)** IS 1.5334 1.8567 AS 1.3129 1.2487 MP 3.5540 1.0984 G 1.5217 2.3922 r * 8.1636 4.4756 Note: the estimation method is GMM. Standard errors are in parentheses. *, **, *** - indicate significance at the 10%, 5% and 1% significance level, respectively. k=0 and l=+1 for both Macedonia and the Slovak Republic, j is indicated in the relevant row. Starting with the estimated IS equations, we could see that the output gap process is more backward- than forward-looking in Macedonia as implied by the estimate of y : 0.48. The same cannot be said about the output gap process in the Slovak Republic where y was estimated at 0.65 implying more weight on the expected output gap in the process. Given the micro-foundations of the model, this would suggest that consumption habit formation is much stronger in Macedonia than the Slovak Republic. More broadly, it could also suggest that adjustments of the capital stock are more costly in Macedonia than in the Slovak Republic. Further, the strength of the interest rate (credit) transmission channel of monetary policy is almost three times higher in the Slovak Republic than Macedonia where the 1 estimates are 0.01 and 0.04, respectively. This suggests that the Slovakian economy reacts much stronger to a given change in monetary policy vis-à-vis Macedonia, so that the economy is easier to stabilize using interest rate adjustments, 10 ceteris paribus. Further, the estimated elasticity of output to changes in the real effective exchange rate, 2 , is about two times higher in the Slovak Republic than Macedonia, where the estimates are 0.06 and 0.12, respectively. The relative magnitudes of 1 and 2 in both economies imply that stabilization of output could be more effective through the exchange rate transmission channel rather than the interest rate transmission channel of monetary policy. Under the exchange rate targeting regime, however, this channel is much less effective because most of the changes in the real effective exchange rate are bound to happen through the adjustment in relative prices (i.e. domestic relative to foreign). Since prices are usually less flexible than exchange rates, macroeconomic stabilization of output using the exchange rate channel would result in longer-lasting deviations of output from its potential, other things equal. The estimates of 3 for Macedonia and the Slovak Republic are 0.23 and 0.14, respectively, implying that a larger portion of consumers in Macedonia relative to the Slovak Republic are rule-of- thumb consumers, i.e. non-Ricardian households (see Gali et al., 2007). Overall, the IS equation fits somewhat better the data of the Slovak Republic than Macedonia where the adjusted R squares are 0.54 and 0.52, respectively. Nevertheless, the estimates of IS for the Slovak Republic and Macedonia, of 1.85 and 1.53, imply that it is the Slovak Republic who faces marginally higher aggregate demand shocks. Consider now the estimated AS equations. In order to achieve a reasonable fit to the CPI inflation data we had to use long-differencing, namely year-to-year changes in CPI as the dependent variable, and the annual cumulative output gap and year-to-year change in the nominal effective exchange rate as the explanatory variables.9 The estimates of for the two countries suggest that the inflation dynamics is driven more by inflation expectations rather than past inflation. It appears that the estimate of = 0.57 applies to both Macedonia and the Slovak Republic. The elasticity of inflation to an increasing output gap, 1 , is estimated to be almost three times higher in Macedonia, 0.039 versus 0.015 in Slovakia. The relative magnitudes of 1 thus imply about two times higher output rigidity in Macedonia. Hence, a given increase in capacity utilization 9 Note that the year-to-year long differencing on quarterly data induces by definition third-order autocorrelation in the AS shocks that is handled according in estimation of the HAC covariance matrix. 11 (demand pressure) will be translated in to a three times higher increase in marginal costs in Macedonia relative to Slovakia. In other words, the flexibility of the potential output to adapt to the increased demand is much lower in Macedonia. The elasticity of CPI inflation to changes in the nominal effective exchange rate, 2 , is estimated to be 0.024 and 0.037 for Macedonia and the Slovak Republic, respectively. The about 65 percent lower exchange rate pass-through in Macedonia could imply either higher monopolistic power of the importers, or a higher share of services in final sales of imports in Macedonia. Further, the second-round effect of CPI inflation on wages as a result of an exchange rate increase could be higher in the Slovak Republic, which would further enforce the exchange rate pass-through. Overall, the AS curve fits the data of the Slovak Republic better than those of Macedonia with the respective adjusted R-squares being 0.86 and 0.78. Based on the relative estimates of the standard deviation of AS shocks AS , Macedonia appears to be subject to somewhat higher supply shocks. When estimating the MP reaction function for NBRM we have included a dummy variable to account for the two spikes in the NBRM interest rate series in 2001Q2 and 2002Q4.10 The estimate of the coefficient is 3.76 for Macedonia which satisfies the stability conditions (the model has a stable solution), see also Benigno et al. (2007). The estimates of the MP rule for Slovakia imply that the SNB smoothes the interest rate ­ where i is estimated to be 0.64 ­ increases interest rate in response to increasing expected inflation, with =1.02, and depreciation of the Slovak koruna, = 0.40 . Overall, the MP equation fits better the data of the Slovak Republic, and only with the use of a dummy variable for the two spikes in the NBRM interest rate series the adjusted R squares for Macedonia and the Slovak Republic turned out to be 0.70 and 0.46, respectively. Nevertheless, the MP discretionary shock within the context of the presented MP rules is significantly higher in Macedonia, with the standard deviation of 3.55, than in the Slovak Republic; estimated standard deviation of 1.10. 10The spike in 2001Q2 was due to the domestic conflict and the Central bank sterilization of the surge in government defense related expenditures. The situation stabilized by early 2002, but in 2002Q4 the Central Bank had to intervene again because the government went on a spending spree just before the September 2002 elections. Once the government changed, the new authorities immediately entered into a stand-by agreement with the IMF and tightened both fiscal and monetary policy considerably. 12 As discussed before, changes in the government expenditure relative to GDP are described by an AR(1) process for both countries. The AR coefficients in the two countries are estimated to be negative but differ in their magnitude and significance. Namely, the estimates of g for Macedonia and the Slovak Republic are -0.09 and -0.33, respectively. The estimate of negative autocorrelation of changes in government consumption in Macedonia is insignificant and thus not different from zero ­ implying that the changes in government consumption in Macedonia follow the white-noise process and that the fiscal stance in this respect is relatively stable in Macedonia. On the other hand, there seems to be significant negative autocorrelation in government consumption-per-GDP changes in Slovakia implying frequent changes in the fiscal stance in Slovakia. Furthermore, the magnitude of the estimated fiscal (government consumption) shock is higher for Slovakia than Macedonia, 2.17 and 1.52, respectively. This further reinforces the relatively higher uncertainty about future economic environment in Slovakia induced by a changing share of government consumption in GDP. Finally, the shock to the real effective exchange rate, as implied by the UIP condition has an estimated standard deviation of 8.16 and 4.48 for Macedonia and the Slovak Republic, respectively. This implies that Macedonia's economy is exposed to about two times larger external shocks compared with the Slovak Republic. The exchange rate shocks could comprise e.g. changes in the terms of trade, the effective foreign real interest rate, or changes in capital flows. 5. Microeconomic Narrative of the Macroeconomic Estimates In this section we focus on justifying the differences in selected estimates of the model structural coefficients for Macedonia and the Slovak Republic using microeconomic evidence. We focus on the coefficient estimates in the IS and AS curves because these equation are richest in the behavioral characteristics of the economies and represent the most important constraints for maximizing monetary and fiscal policy objectives. Namely, from the estimates of the IS curve we focus on illustrating why in Macedonia (i) the consumption habit formation or cost of capital stock adjustments could be higher, (ii) the credit channel of monetary policy could be weaker, (iii) the exchange rate effect on 13 net exports could be weaker, and (iv) the portion of the rule-of-thumb consumers (non- Ricardian households) could be higher, vis-à-vis the Slovak Republic. Similarly, we try to illustrate why in Macedonia, relative to the Slovak Republic, (v) the effect of increasing capacity utilization (demand pressure) on inflation could be higher, and (vi) the exchange rate pass-through could be lower. 5.1. Microeconomic Narrative for the IS Curve Estimates (i) Why could consumption habit formation or the cost of capital stock adjustments be higher in Macedonia? Data on CPI's weights, which are typically based on households' consumption patterns, seem to confirm stronger habit formation in Macedonia. The CPI basket changed in both countries between 1999 and 2007; however, the overall change in the structure of the CPI (approximated by the standard deviation of the changes in the weights of the various COICOP categories) was greater in the Slovak Republic. Faster rising incomes in Slovakia compared to Macedonia meant that Slovaks increased spending on non- necessities more than Macedonians. In addition, developments in prices differed considerably; for example, price-adjustment in energy and utilities to cost recovery levels in the Slovak Republic occurred much faster compared to Macedonia, meaning that Slovaks' consumption habits had to adjust much earlier. Table 2: Changes CPI weights in the Slovak Republic and Macedonia Slovak Republic Macedonia 1999 2007 change 1999 2007 change Food and non-alcoholic beverages 293.4 168.3 -125.1 488.2 373.0 -115.2 Alcoholic beverages, tobacco and narcotics 46.3 55.7 9.4 43.0 40.0 -3.0 Clothing and footwear 120.6 42.0 -78.6 79.0 76.0 -3.0 Housing, water, electricity, gas and other fuels 138.8 227.1 88.3 118.1 121.0 2.9 Furnishings, household equipment and routine maintenance of the house 61.3 58.2 -3.1 38.2 50.0 11.8 Health 14.5 36.7 22.3 22.6 34.0 11.4 Transport 98.9 102.7 3.9 90.9 91.0 0.1 Communications 15.9 44.0 28.1 30.5 64.0 33.5 Recreation and culture 95.1 85.1 -10.0 33.0 43.0 10.0 Education 11.6 17.5 5.9 1.1 8.0 6.9 Restaurants and hotels 52.1 87.5 35.4 9.8 49.0 39.2 Miscellaneous goods and services 51.7 75.2 23.5 45.6 51.0 5.4 Standard deviation 54.7 38.6 14 Note: Components sum to 1,000, according to COICOP methodology. Source: for Slovakia Eurostat; for Macedonia State Statistics Office (the Macedonia numbers are the 2002 published numbers. Given that weights were not changed between 1999 and 2005, the 2002 weights are assumed to be valid for 1999. More broadly, higher output gap persistence (see the estimated IS curve in Table 1) also suggests that adjustments of the capital stock are more costly in Macedonia than in the Slovak Republic. The difficulties of Macedonian businesses to take advantage of changing business conditions are discussed under "output rigidity" in Section (v) below. (ii) Why could the credit channel of monetary policy be weaker in Macedonia? The IS curve estimates suggest that Macedonia's output gap is less responsive to interest rate changes than Slovakia's output gap.11 The direct interest rate channel works largely through its impact on domestic credit and hence aggregate demand. Its efficiency is thus crucially dependent on the transmission of changes in the monetary policy rate to deposit and lending rates. This transmission, see Figure 1, can be less effective for several reasons. Figure 1: Lending, Deposit and Monetary Policy Rates in Slovakia and Macedonia 25.0 25.0 20.0 20.0 15.0 15.0 10.0 10.0 5.0 5.0 0.0 0.0 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 Lending rate Deposit rate Lending rate Deposit rate Monetary Policy rate Monetary Policy rate Source: IFS and NBRM; the monetary policy rate is approximated by the CB bill rate for Macedonia and the 3-month interbank rate for Slovakia, see the data description section for details. 11A recent NBRM report on the effect of the interest rate transmission mechanism largely supports the view that this channel is weak. Available in Macedonian at: http://www.nbrm.gov.mk/default- MK.asp?ItemID=98BEA8015487194DB5825340B5C5ECCF 15 First, there is relatively lower competition in the financial sector of Macedonia compared to that of Slovakia. Though Macedonia has a higher number of banks per capita, the Macedonian banking sector is dominated by few large banks preventing competitive pressures to force increased lending and deposits, and to reduce interest rate margins. As a result, the lending rate and the interest rate margins in Macedonia have been higher compared to those in Slovakia (see Figure 1). Slovak banks, facing higher competition borrowed extensively funds from abroad. On the other hand, Macedonian banks, facing limited competition still remain net international creditors despite higher domestic interest rates and profit margins relative to Slovakia (see Table 3). Table 3: Net External Position and Competitiveness of the Banking Sector 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 Macedonia Net external position1 2.4 3.9 5.5 13.0 9.3 10.2 10.8 7.7 7.7 4.7 Banking sector reforms2 2.7 2.7 2.7 2.7 2.7 2.7 2.7 2.7 2.7 2.7 Banking Efficiency3 ... ... 3.3 3.6 3.5 3.5 3.6 4.8 ... ... Slovakia Net external position1 5.2 4.7 8.3 7.3 2.3 -2.5 -6.9 -14.6 -5.7 -9.9 Banking sector reforms2 2.3 2.3 2.3 2.3 2.3 2.7 2.7 2.7 3.0 3.0 Banking Efficiency3 ... ... 4.8 4.4 4.9 4.8 4.8 4.9 ... ... Note: 1/ Net of banking sectors' external claims and liabilities; 2/ EBRD Banking reforms and interest rate liberalization index; 3/ World Bank Financial development indicator; Source: IMF International Financial Statistics, EBRD, World Bank. Second, the mix of macroeconomic policies, namely the de-facto fixed exchange rate and frequent fiscal surprises, have allowed Macedonian banks to enjoy higher risk- free returns (through FX arbitrage) relative to Slovak banks, thus further crowding out lending to the private sector and limiting banks' competition (see Table 3 and Figure 2). Since their introduction in 2004, the average interest rate on the 3-month T-bills issued by the Macedonian Ministry of Finance was 7.5%, compared to 4.1% in the Slovak Republic during the same period. The real (CPI-adjusted) risk-free return was thus, on average, significantly positive (6.2) for Macedonian banks since 2004, while it was negative (-0.3) for Slovakian banks. This could also reduce the relative incentives of Macedonian banks to lend to the private sector. Third, the extent of the involvement of the private sector in financial operations, especially borrowing, is important. Credit to the private sector barely reached 40% of 16 GDP in Macedonia by 2008, compared to more than 50% of GDP in Slovakia (see Figure 2 panel 1). Especially in the earlier years covered in this paper, a large part of the Macedonian economy had relatively limited access to financing due to less competitive banking sector, while those with access to financing were financed at unreasonably high costs. Figure 2: Financial Deepening and Access to and Cost of financing as a constraint of doing business Panel A: Domestic Credit as % of GDP Panel B: Access and Cost of financing as a constraint of doing business 70 2002 60 a ki 50 vaolS 2005 GDPfo 40 %ni 30 a 2002 20 ni 10 doeca M 2005 0 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 0% 20% 40% 60% 80% Slovakia Macedonia Access to financing Cost of financing Source: IMF IFS Source: BEEPS 2002 and 2005. Fourth, the extent of the euro-ization (formerly dollarization) is much larger in Macedonia than Slovakia, making the actual lending/deposit base affected by the changes in the monetary policy rate on local currency denominated instruments much narrower, and the changes in the monetary policy rate thus less effective in managing aggregate demand. At the end of 2006, only 1.5% of loans to households and 33% of loans to the corporate sector were extended in foreign currency in Slovakia, compared to 45% of loans to the household sector in Macedonia and 58% of loans to the corporate sector. (iii) Why could the exchange rate effect on net exports be weaker in Macedonia? Several factors could explain the weaker output response to changes in the real exchange rate in Macedonia compared to the Slovak Republic that we have estimated. 17 First, the structure of Slovak imports contains a larger share of goods expected to have higher import elasticities. Generally, import demand elasticities for manufactured goods tend to be higher compared to those of raw materials and food,12 which can explain the lower responsiveness of imports to REER development in Macedonia. Machinery and equipment account for the largest share of Slovak imports (see Table 4). On the other hand, Macedonian imports are dominated by goods imported for processing and re-exported (textiles and iron and steel; the bulk of imports classified as other manufactured products) as well as a sizable share of basic consumption goods (food and oil). Imports for processing are more sensitive to wage pressures (rather than REER developments) which have been modest due to the high unemployment in Macedonia. Table 4: Imports and exports structures, SITC classification, 1999-2006, in % of total Imports Structure Exports Structure Slovakia 1999 2003 2006 1999 2003 2006 Food, drinks and tobacco 6.2 4.3 4.8 3.5 2.8 3.8 Raw materials 4.0 3.4 3.3 4.0 2.5 2.4 Fuels, lubricants and related materials 13.0 12.1 14.1 4.7 5.1 6.4 Chemicals and related products 11.3 9.8 8.8 7.8 5.0 5.5 Other manufactured products 27.8 29.1 29.9 40.4 36.4 32.8 Machinery and transport equipment 37.7 40.7 38.8 39.4 47.4 48.5 Macedonia Food, drinks and tobacco 13.8 12.8 10.7 19.0 16.8 16.0 Raw materials 3.2 2.6 3.6 4.3 2.9 4.7 Fuels, lubricants and related materials 9.1 14.0 20.2 1.9 5.4 9.4 Chemicals and related products 10.4 11.1 9.7 4.6 5.2 4.2 Other manufactured products 20.8 20.0 36.7 61.0 63.6 60.7 Machinery and transport equipment 20.0 18.8 18.3 7.0 5.9 4.9 Source: EUROSTAT for Slovakia and State Statistics Office for Macedonia Second, Macedonian exports are concentrated in only few sectors (firms) with low excess capacity and growing competition. Almost 50% of Macedonian exports are re-exports of iron and steel, and textiles, or, in recent years, fuel where the corresponding firms operate close to full capacity utilization. As a result, Macedonian reliance on few export companies has grown over time resulting in growing export concentration. On the other hand, there are relatively more exporting Slovak firms, and Slovakia's exports 12Among the first who made this observation was Kreinin (1973). 18 cover a greater variety of products and more markets. Hence, Slovakia's exports are less reliant on few key industries, which enables firms to take advantages of changes in their external price competitiveness. Table 5 below presents developments in export concentration in Macedonia and the Slovak Republic. Although Slovakia increased concentration of its exports by products, it has diversified exports across more destinations. On the other hand, Macedonia exports become less diversified in both product and destinations. Table 5: Developments in export concentration, in percent of total exports Slovakia 1996 2006 Macedonia 1996 2006 top 5 countries 70.2 56.0 top 5 countries 62.4 69.2 top 10 countries 85.3 77.5 top 10 countries 80.7 85.1 top 5 products 19.6 37.6 top 5 products 36.9 45.3 top 10 products 30.6 46.2 top 10 products 54.7 61.0 Source: COMTRADE Third, Macedonian exports had a more restricted market access due to lower trade integration into the European region, low levels of FDI, and inability to meet quality standards. Compared to Slovakia, which gained duty free access to the EU market under the Europe Agreement in 1995, Macedonia's access to the EU market occurred much later with the signing of the Stabilization and Association Agreement in 2001, and more intensive integration with the immediate neighborhood started only with the 2006 CEFTA agreement. The FDI inflow, typically associated with increased market access and transfer of know-how,13 has been much lower for Macedonia compared to Slovakia. Over the 1997-2007 decade, FDI averaged 4.6% of GDP in Macedonia, compared to 6.4% of GDP in the Slovak Republic; as a result, cumulative inflows of FDI over this period in Macedonia reached USD2.3 billion, or only 10% of the Slovak USD23.7 billion inflow. Also, greater compatibility with quality standards can explain the greater responsiveness of Slovak exports to changing price competitiveness. According to the International Organization for Standardization, only 217 Macedonian firms were ISO 9001-certified at the end of 2006 compared to 2195 of Slovak firms. Similar 13This link has been well-demonstrated especially by the Volkswagen FDI in the Slovak Republic. 19 discrepancies emerge when comparing certification on other quality standards (see Table 6).14 Table 6: Number of ISO certified firms Dec.01 Dec.02 Dec.03 Dec.04 Dec.05 Dec.06 ISO 9001:2000 Slovakia 144 768 1,148 2,008 2,050 2,195 Macedonia 1 7 47 133 154 217 ISO 14001 Slovakia 73 70 165 184 222 305 Macedonia 1 1 5 6 8 Source: International Standards Organization 2006 Survey Finally, the sensitivity of net exports to changes in REER may be qualitatively different due to the different exchange rate regime applied in the two countries. Under a flexible exchange rate regime (the Slovakia's case) the exchange rate adjusts to the domestic economic activity and external environment, while under an exchange rate peg (the Macedonia's case), the domestic economic activity and prices have to adjust to the external developments ­ especially in the case of small open economies. Therefore, under the peg, changes are bound to happen mostly through adjustments in prices. This will obviously change relative prices in the economy and hence the production structure, because price adjustments will vary across different sectors due to varying price rigidities and export orientation. Hence, also the pegged exchange rate contributes to the explanation of the relatively lower output gap elasticity to exchange rate changes. (iv) Why could the portion of the rule-of-thumb consumers (non-Ricardian households) be relatively higher in Macedonia? Ricardian households will increase their savings in face of increased government consumption (spending) to retain funds for accommodating future increases in taxes. They thus buy government bonds in response to fiscal impulse so that there is no actual effect on aggregate demand. The presence of significant portion of non-Ricardian 14The inability to meet quality standards has translated into growing reliance on low-value added exports in Macedonia reflected in deteriorating (at best not-improving) terms-of-trade (2% decline over 2000-2005). On the other hand, Slovakia has witnessed a considerable improvement in its terms-of-trade, 23.3% over 2000-2005. 20 households in transition economies should not be a surprise, given the relatively stringent conditions needed for Ricardian equivalence to hold: perfect capital markets, consumption smoothing behavior, intergenerational concerns, and no distorting effects of taxes, among others (also see Barro 1974; Briotti, 2005). We look into selected aspects of the non-Ricardian behavior in Macedonia in comparison with Slovakia. First, less efficient and complete financial markets in Macedonia relative to Slovakia reduce the availability of financial instruments to offset the anticipated future changes in fiscal policy (see Table 7). Poor competition in the banking sector, underdeveloped government debt markets, limited financial deepening and sophistication, and external capital restrictions might have reduced incentives to save rather than consume the extra household income as a result of increased government spending, and increased liquidity constraints which prevent effective consumption smoothing. Table 7: Capital market development indicators 2003 2004 2005 2006 2003 2004 2005 2006 Slovakia Macedonia as % of GDP Stock market capitalization 8.4 10.5 9.3 10.1 7.8 7.7 11.1 17.7 Government securities, outstanding 32.5 32.1 28.0 26.3 14.4 13.6 16.1 15.7 S&P Rating A- A- A A BB BB BB+ BB+ Non-bank financial institutions reforms, EBRD index 2.7 2.7 2.7 3.0 1.7 2.0 2.0 2.3 Source: Stock market capitalization from WDI, Government securities outstanding ­ author calculations based on the Ministry of Finance and Central bank data, S&P rating from Standard&Poor's sovereign ratings; Index of reform of non-bank financial institutions from EBRD Transition Indicators. Second, poverty is more prevalent in Macedonia, making the affected people consume more or all of their disposable incomes while leaving less of it for savings. Around a fifth of Macedonia population is absolutely poor (World Bank, 2005). Since comparable poverty data for Slovakia are not readily available, an indirect illustration can be drawn from relative per capita incomes and income inequality. Namely, in 2006, Macedonia's per capita GNI at PPP was USD7850, significantly below Slovak's GNI per capita at PPP of USD17060, also the GINI indexes were 39.0 in 2003 for Macedonia, and 25.8 in 1996 for Slovakia.15 15World Bank estimates "Distribution of Income or consumption 2.7". More recent data are not available. 21 Third, a history of more prudent fiscal policies and smaller government could explain the larger portion of non-Ricardian households in Macedonia, as many households have not felt yet significant adverse effects due to excessive government consumption. With the exception of the 2001-2002 period, the Macedonian budget was largely balanced over the last decade (compared to an average deficit of around 5% of GDP in Slovakia), i.e. changes in government consumption were compensated by changes in current tax revenues, and the government size in Macedonia has been significantly smaller compared to Slovakia ­ government expenditures averaged around 37% and 44% of GDP over 1997-2007 in Macedonia and the Slovak Republic, respectively. 5.2. Microeconomic Narrative for the AS Curve Estimates (v) Why could the effect of increasing capacity utilization (demand pressure) on inflation be higher in Macedonia? The Phillips curve estimates imply relatively higher output rigidity in Macedonia compared with Slovakia. This means that Macedonian producers are unable to respond to higher demand without considerably increasing costs and thus hurting their competitiveness. There are several competing explanations for the capacity constraints of Macedonian businesses. Macedonian producers may be operating at higher levels of capacity utilization. Given the depreciated and largely obsolete inherited capital stock and low investments over the last 15 years (averaging 20% of GDP compared to Slovakia's 29% of GDP) it is reasonable to expect that Macedonian producers have lower opportunities to expand output within existing facilities at relatively low marginal costs. However, data on capacity utilization in manufacturing from the Eurostat and the Macedonian statistical office are not supportive of this argument. Capacity utilization in manufacturing fluctuated between 65-70%16 during 2005-2008 in Macedonia compared 16Business Tendencies in the Manufacturing Sector, Macedonia State Statistics Office, available at: http://www.stat.gov.mk/statistiki.asp?ss=11.03&rbs=1 22 to 70-80%17 in the Slovak Republic. However, it should be taken into account that these surveys are most likely not fully comparable and that access capacity may be located largely in sectors for which there is limited demand. This brings us to a more likely explanation of higher output rigidity in Macedonia, the gap between domestic demand and production, in other words, narrow production base. Around 60% of Macedonian industrial production is accounted for by sectors with rigid capacity constraints (electricity, oil) or export-oriented sectors which operate at almost full capacity (iron and steel, and textiles). Further, the size and structure of both economies differ considerably. While Macedonian economy is dominated by small (even micro) entities, most of the value added in the Slovak Republic is generated by large enterprises (see Table 8) which probably have sufficient excess capacity or greater ability to expand production with limited cost implications (see also Figure 2 panel 2), and better exploit the economies of scale. Table 8: Structure of economies by size of entities (as % of total) Number of entities Number of employees Value added Small Medium Large Small Medium Large Small Medium Large Slovakia 93.1 5.5 1.4 28.7 22.5 48.8 25.1 17.3 57.6 Macedonia 99.0 0.6 0.4 63.8 13.6 22.6 47.5 13.0 39.5 Source: For Slovakia: European Business, Facts and Figures, 2007 Edition - Eurostat Statistical book (p.30), data refers to 2004, for Macedonia: author calculations based on National Accounts data from the State Statistics Office. Note: Data refers to non-financial business economy A considerably more favorable business environment enables Slovak businesses to react faster and cheaper to increased demand. The Slovak Republic ranks 32nd on the 2008 Doing Business indicator compared to Macedonia's 75th ranking (despite a considerable improvement from 2007, and from before when the country ranked around 100th). Overall, the better business environment provides Slovak businesses with more and cheaper financing, less corrupt and faster public administration, and more efficient 17Current level of Capacity utilization in manufacturing industry, EUROSTAT, available at: http://epp.eurostat.ec.europa.eu/tgm/table.do?tab=table&init=1&plugin=1&language=en&pcode=bs070 23 protection of creditor and property rights compared to their Macedonian counterparts (see Table 9). Table 9: Ranking on 2008 Doing Business indicators SR MK SR MK Doing Business 32 75 Getting Credit 7 48 Protecting Investors 98 83 Starting a Business 72 21 Paying Taxes 122 99 Dealing with Licenses 50 76 Trading Across Borders 90 72 Employing Workers 75 128 Enforcing Contracts 50 84 Registering Property 5 91 Closing a Business 36 127 In addition, a better skills structure of the Slovak labor force could explain the ability of Slovak firms to expand output faster and cheaper. Even though Macedonia has a much higher unemployment rate (standing at around 34% in 2007, compared to around 13%18 in Slovakia), an unemployed Macedonian is much less skilled and educated compared to his counterpart in Slovakia. Around 55% of the unemployed in Macedonia were first-time job seekers (considerable number aged 35 and above) compared to only 25% in the Slovak Republic. Above 40% of unemployed in Macedonia have less than secondary education compared to around 28% in the Slovak Republic. This suggests that for Macedonian businesses it is more difficult and expensive to find and hire skills that are in demand compared to Slovak businesses. This is also reflected in the relatively lower score in the Doing Business survey concerning the legal framework governing employment of workers (see Table 9, Employing Workers). (vi) Why could the exchange rate pass-through be lower in Macedonia? The estimate that Slovak CPI reflects exchange rate changes more than Slovak CPI can be explained by differences in functioning of product markets as well as trade integration of the two countries. Functioning competitive product markets imply that changing prices reflect developments in costs. Administered prices can break this link. Most of the prices are set liberally in both economies; however, the small set of administered prices in Macedonia 18Data is for 2006 24 (utilities, electricity) has shown greater resistance to economic developments (largely reflecting the authorities' policies to protect the living standard of the population). End- prices of electricity increased five-fold between 1997 and 2007 in the Slovak Republic compared to only doubling in Macedonia over the same period. Even though prices of smaller number of goods are being regulated in Macedonia compared to the Slovak Republic; the share of administered prices in the CPI in 2006 was 1.2% and 23.4% in Macedonia and Slovakia (EBRD, 2007), the problem appears to lie in the commercial orientation of the provision of major CPI basket articles. Additionally, constraints to market entry, based on either the market size or institutional arrangement, can reduce the exchange rate pass-through by allowing monopolistic importers to charge excessive profit margins on their products and adjust prices in an asymmetric way. Equally low competition in the domestic market then supports the survival of such monopolistic structures in the import industry. The lower degree of market competition in Macedonia compared to Slovakia can explain the lower sensitivity of CPI to exchange rate changes. Namely, the Slovak Republic scored 3.3 on EBRD's index of competition policy19 in 2007 and 0.39 on OECD's Index of pro- competitive reforms20 (data is for 2005), compared to Macedonia's score of 2.3 on the EBRD index and 0.59 on the OECD index. Finally, the differing exchange rate pass-through may reflect differences in the living standards of the population in both countries (i.e. the construction of the CPI) as well as differences in the extent of trade integration. Macedonian CPI gives a considerably larger weight to food and beverages (around 40% compared to Slovakia's around 20%) which are mostly domestically produced and as a result shielded from exchange rate developments (see Table 10). Also, imports of goods account for above 80% of Slovak GDP compared to only around 60% of Macedonian GDP. Further more, imports of goods which are closely reflected in the CPI (foods and beverage, fuels and lubricants, transport equipment and consumption goods) account for 35.7% of GDP in the 19The index ranges from 1 to 4+ with 1 referring to "No competition legislation and institutions" and 4+ referring to "Standards and performance typical of advanced industrial economies; effective enforcement of competition policy; unrestricted entry to most markets" (EBRD, 2007, p.211). Most of the advanced transition economies have an Index of competition policy of 3.3, with only Estonia performing better (Index of competition policy of 3.7) 20The index range is between 0 and 1 with smaller value implying better performance on the index. Index data from Miroudut et al. (2007). 25 Slovak Republic in 2006, compared to 28.4% of GDP in Macedonia, suggesting the imports have also a larger share in Slovak consumption. Table 10: Import structure for Macedonia and the Slovak Republic Macedonia Slovak Republic 2005 2006 2005 2006 Food and beverages 6.3 6.3 3.6 3.6 Industrial supplies nes 22.7 24.6 22.1 23.6 Fuels and lubricants 10.3 11.8 10.0 10.8 Capital goods 6.5 6.9 17.5 20.9 Transport equipment 3.4 4.1 11.5 13.3 Consumption goods nes 6.1 6.1 7.5 7.9 Goods nes 0.0 0.1 0.5 0.0 TOTAL 55.4 59.9 72.6 80.2 Included closely in CPI 26.2 28.4 32.5 35.7 Source: COMTRADE, Imports according to BEC methodology 6. Impulse Response Analysis The impact of various shocks on the main variables of interest is commonly analyzed using impulse response functions (IRFs). An IRF gives an answer to a question of what will happen to an endogenous variable of a given economic system when the system is shocked by a certain event, such as aggregate demand shock21, while keeping other things constant. The endogenous system is then rolled forward so that in the long run the impulse response is expected to converge back to the equilibrium (the steady state) in response to transitory shocks. We carry out the impulse response analysis using the estimated model for Macedonia and the Slovak Republic and compare the IRFs of the two countries for selected variables. In order to recover IRFs, the linearized rational expectation model is put into state-space form and solved using the QZ solution algorithm of Sims (2002). The solved model has then a VAR structure readily allowing the computation of the IRFs. We focus on the responses of the output gap, inflation, interest rate and the real exchange rate to the aggregate demand, aggregate supply, monetary policy, fiscal and external (exchange rate and foreign) shocks. The estimated IRFs are presented in Figure 3 and 21The aggregate demand shock could be result of e.g. the change in consumers' preferences. 26 Figure 4. All impulse responses are to one standard deviation of a shock.22 Figure 3: Responses of the output gap, inflation, interest rate and exchange rate to demand (IS), supply (AS) and monetary policy (MP) shocks in Macedonia (MK) and the Slovak Republic (SR). The first row of Figure 3 shows the impulse responses of the output gap, inflation, interest rate, and real exchange rate to a positive demand (IS) shock. We can observe that at impact the domestic IS shock has slightly higher impact on the output gap in Slovakia ­ partly because its size is bigger (see Table 1). However, the adjustment back to the steady state is faster in Slovakia perhaps due to more forward-looking nature of the process driving output gap formation (viz. Table 1). As the output gaps in both countries 22This means that the relative size of the shocks hitting Macedonia and Slovakia may differ for each type of shock. Alternatively, one could hit both systems with shocks of the same size for both countries and for each type of shock. This will however disregard the fact that the two countries face shocks of different sizes and that even a bigger shock can have a smaller immediate impact on variables as the structural form is solved into the reduced form, used for the IRFs, where the reduced-from coefficients are a non-linear function of the structural coefficients including the estimated standard deviations. We thus prefer using one standard deviation shocks in the IRF analysis. 27 open, the higher sensitivity of inflation to excess demand in Macedonia (see Table 1) results in about three times a higher impact of the IS shock on inflation in Macedonia compared to Slovakia, where in addition the adjustment back to the equilibrium is much longer in Macedonia. This is partly due to the fact that the reaction of Macedonia's interest rate to the IS shock is negligible compared to Slovakia, where the monetary authority react to increasing inflation by raising interest rates. The Macedonian denar appreciates sharply in effective terms as inflation rises, while the response of the real exchange rate (appreciation of the Slovakian koruna in effective terms) is proportionally smaller in Slovakia due to a more subdued inflation response. Overall, Slovakia's economy appears to be much efficient in absorbing the IS shock. The second row of Figure 3 presents the variables' responses to a positive supply (AS) shock. Once the AS shock hits the economy, inflation increases at impact. The adjustment of inflation back to the steady state appears to be faster in Slovakia, while Macedonia experiences a short and minor deflation period before inflation stabilizes at its equilibrium value. The Slovakian central bank reacts to rising inflation by increasing its interest rate. On the other hand, the Macedonian central bank does not respond to the domestic AS shock. Nevertheless, both the denar and the koruna appreciate in effective terms, where a much larger appreciation is needed in Macedonia to stabilize the economy, given the virtually zero response of the Macedonian interest rate to the domestic AS shock. As a result of an increasing interest rate and appreciation of the koruna, Slovakia's output gap declines more than the Macedonia's output gap in response to the domestic AS shock. Given the lower output gap response, one may argue that the AS shock is somewhat better absorbed by the Macedonian economy. The third row of Figure 3 displays the variables' responses to a positive monetary policy (MP) shock ­ a discretionary increase in the interest rate. As seen in the third panel, the MP discretionary shock is much higher in Macedonia than Slovakia (see also Table 1). Nevertheless, in both countries the domestic MP shock has only very short- lived effect on the interest rate. As the interest rate increases, the exchange rate responds proportionally, so that the Macedonian denar appreciates significantly more in effective terms than the Slovakian koruna. The raising interest rate and appreciating domestic currency decrease the output gap in both countries, where the initial output gap response 28 is more pronounced in Slovakia. However, the output gap in Slovakia swings into positive territory before stabilizing around its steady state. Inflation too declines in response to domestic currency appreciation and an increased domestic interest rate in both countries, with a higher decline in Slovakia. Despite the higher size of the MP shock in Macedonia, the economy appears to adjust faster to its equilibrium (steady state) than the Slovakian economy does to a corresponding domestic MP shock. Figure 4: Responses of the output gap, inflation, interest rate and exchange rate to government consumption (FP), and external (ER) shocks in Macedonia (MK) and the Slovak Republic (SR). The first row of Figure 2 shows the responses of the variables to a positive fiscal shock -- here an idiosyncratic increase in the government consumption-to-GDP ratio. One can see in the first panel that as the positive fiscal shock hits the economy the output gap response in a positive way in both countries, where in Macedonia the response is about six times larger, despite the standard deviation of the Slovakian fiscal shock being almost two time higher than that of the Macedonian fiscal shock (see Table 1). As the output gap 29 opens, also inflation increases in both countries in response to a positive fiscal shock, where again the increase in Macedonia's inflation is much higher, about eight times, than the response of inflation in Slovakia. The response of the domestic interest rate in Macedonia to the fiscal shock is negligible and essentially zero, as the monetary policy in Macedonia does not react to this impulse. In contrast, the Slovakian interest rate increases proportionally to the rising output gap and inflation, which, once deviating from its steady state, makes the Slovak National Bank react. The responses of the domestic variables are then reflected in the real exchange rate's response in the two countries. While the exchange rate's response in Slovakia is relatively subdued proportionally to the responses of the domestic variables, the exchange rate's response in Macedonia shows significant appreciation of the denar in effective terms, especially due to the relatively high increase in CPI inflation. Overall, the Slovakian economy seems to be coping with the fiscal shocks better than the Macedonian economy. The second row of Figure 2 displays the variables' responses to an external shock, equivalent to an idiosyncratic depreciation of the domestic currency in effective terms, i.e. an increase in the real effective exchange rate. As the negative external shock hits the economy the Macedonian denar and Slovakian koruna depreciate in real effective terms, where the initial depreciation is much more pronounced in Slovakia. This is partly due to the exchange rate targeting nature of the monetary policy in Macedonia, where the National Bank of the Republic of Macedonia reacts immediately and increases its interest rate to stabilize the nominal exchange rate. This also helps in containing the real effective exchange rate's response. On the other hand, the Slovak National Bank reacts to the depreciation of the koruna in real effective terms only to the extent this depreciation is affecting output and inflation. The output gap and inflation shoot up at the impact of the koruna's depreciation, so that the Slovak National Bank increases the domestic interest rate. After the interest rate increase takes effect it makes the real exchange rate to swing back to the negative territory before it approaches its steady-state value. The same pattern could be observed in the response of the output gap and inflation over time, i.e. after the initial positive response, the pattern changes and the response becomes negative as the interest rate increases and the koruna appreciates, after its initial depreciation. In Macedonia, on the contrary, the initial responses of the output gap and inflation are 30 negative due to the high increase in the domestic interest rate in an effort to stabilize the nominal exchange rate. The responses then faze out and converge to the steady-state values for output and inflation. In general, Macedonia is able to contain the effect of the external shock better than Slovakia, but at the cost of a large spike in the domestic interest rate and significantly negative responses of the output gap and inflation. While allowing for larger volatility in the exchange rate and focusing on inflation targeting Slovakia is able to secure lower volatility of output and inflation in the long run (see Table 11). This is because if the external relative price is not allowed to adjust to absorb negative external and domestic shocks the adjustment has to happen in the domestic economy through prices and production volumes. The latter two are much less flexible than the exchange rate, and should exchange rate targeting be the long-term choice for monetary policy the Macedonian economy has to become more flexible to ensure that macroeconomic stabilization is not in the way of long-run growth. This can be achieved through improved labor markets functioning, financial intermediation, capital stock adjustments, and more competitive production. Table 11: Asymptotic standard deviation of economic variables for Macedonia and Slovakia based on the estimated models' simulations Volatility of Output gap Inflation Interest rate Real exchange rate Macedonia 5.9931 4.6698 8.2882 5.0862 Slovakia 4.3198 3.0333 3.7332 5.2433 7. Conclusion In this paper, an attempt was made to fit a New Keynesian Policy model to the data on Macedonia and Slovakia, while providing some justification for the differences in estimated model parameters and structural shocks using a gap analysis of relevant microeconomic data. Under the current monetary policy regime of exchange rate targeting, Macedonia needs to improve the transmission mechanism of the monetary policy by increasing competition in and efficiency of financial intermediation, promoting financial deepening, increasing access to finance, improving the response of domestic 31 producers to changes in external price competitiveness, eliminating the effect of regulated (subsidized) prices on export concentration while promoting export diversification, and making sure that regulated prices do not significantly distort consumption preferences. In addition, Macedonia has to address especially real rigidities related to costs of production expansion, promoting development of specialized clusters, horizontal and vertical integration of production, fast and adequately priced finance, and ensure overall improvements in investment climate. Based on the differences in the estimated functioning of the Macedonian and Slovak economies and the applied monetary policy regimes in the two countries, we come to a conclusion that small open economies with stronger economic rigidities should apply monetary policy regimes that allow for more flexible adjustments in external relative prices to enhance their macroeconomic stability. In other words, if a small open economy chooses to adopt exchange rate targeting as its monetary policy regime and to give up a portion of its adjustment flexibility to domestic and external shocks, it needs to work extra hard to generate additional flexibility within its production factors' and product markets to avoid higher inflation and output volatility outcomes in their stabilization efforts. 32 References An, S., Schorfheide, F. (2005). Bayesian Analysis of DSGE Models. Discussion Paper 5207, Centre for Economic Policy Research. Barro, R.J. (1974). Are Government Bonds Net Wealth? Journal of Political Economy 6, vol. 82 (Nov. ­ Dec), 1095-1117. Benigno, G., Benigno, P. (2003). Designing Targeting Rules for International Monetary Policy Cooperation. Journal of Monetary Economics vol. 53(3), 473-506. Benigno, G., Benigno, P., Ghironi, F. (2007). Interest Rate Rules for Fixed Exchange Rate Regimes. Journal of Economic Dynamics & Control 31, 2196-2211. Briotti, G.M. (2005). Economic Reactions to Public Finance Consolidation: a Survey of the Literature. European Central Bank Occasional Paper No. 38, (Oct.). Buncic, D., Melecky, M. (2008). An Estimated New Keynesian Policy Model for Australia. Economic Record 264, vol. 84, 1-16. Chinn, M.D., Meredith, G. (2004). Monetary Policy and Long-Horizon Uncovered Interest Parity. IMF Staff Papers 51, 409.430. Christiano, L. J., Eichenbaum, M., Evans, C. (2001). Nominal Rigidities and the Dynamic Effects of a Shock to Monetary Policy. Working Paper 01-07, Federal Reserve Bank of Cleveland. Clarida, R., Gali, J., Gertler, M. (2001). Optimal Monetary Policy in Open versus Closed Economies: An Integrated Approach. American Economic Review 91, 248­52. Clarida, R., Gali, J. and Gertler, M. (2002). A Simple Framework for International Monetary Policy Analysis. Journal of Monetary Economics 49, 879­904. EBRD (2007). Transition Report 2007. European Bank for Restructuring and Development, London. Ferreiraa, A.L., León-Ledesma, M.A. (2007). Does the Real Interest Parity Hypothesis Hold? Evidence for Developed and Emerging Markets. Journal of International Money and Finance 3, Vol. 26, 364-382. Fuhrer, J., Moore, G. (1995). Inflation Persistence. Quarterly Journal of Economics 110 (1), 127-159. Galí, J., Gertler, M. (1999). Inflation Dynamics: A Structural Econometric Analysis. Journal of Monetary Economics 44, 195­222. 33 Galí, J., López-Salido, J.D., Vallés, J. (2007). Understanding the Effects of Government Spending on Consumption. Journal of the European Economic Association vol. 5(1), 227-270. Giordani, P. (2004). Evaluating New-Keynesian Models of a Small Open Economy. Oxford Bulletin of Economics and Statistics 66 (Supplement), 713­733. Iradian, G. (2007). Rapid Growth in Transition Economies: Panel Regression Approach. IMF Working Paper 07/170. Johansen, S. (2005). What is the price of Maximum Likelihood. Paper presented to the Model Evaluation Conference, Oslo, May 2005. Kreinin, M.E. (1973). Disaggregated Import Demand Functions: Further Results. Southern Economic Journal 1, vol. 40 (Jul.), 19-25. Mark, N.C., Moh, Y.K. (2001). What do Interest-Rate Differentials Tell us about the Exchange Rate? Manuscript, Ohio State University. Melecky, M. (2008). A Structural Investigation of Third-Currency Shocks to Bilateral Exchange Rates. International Finance 1, vol.11, 19-48. Miroudut S., Pinali, E., Sauter, N. (2007). The Impact of Pro-Competitive Reforms in Developing Countries. OECD Trade Policy Working Paper no.54. Monacelli, T. (2005). Monetary Policy in a Low Pass-Through Environment. Journal of Money, Credit and Banking 37, 1047.1066. Sims, C.A. (2002). Solving Linear Rational Expectations Models. Computational Economics 20, 1­20. Svensson, L.E.O. (2000). Open-Economy Inflation Targeting. Journal of International Economics 50, 155-183. Taylor, J.B. (1993). Discretion versus Policy Rules in Practice. Carnegie-Rochester Conference Series on Public Policy 39, 195-214. World Bank (2005). Macedonia: Poverty Assessment 2002-2003. World Bank, Washington, DC. 34