A Financial Cycle for Albania

Transcription

1 A Financial Cycle for Albania Vasilika Kota and Arisa Goxhaj (Saqe) FInancial Stability Department Bank of Albania (First draft) The views expressed herein are of the authors and do not necessarily reflect the views of the Bank of Albania.

2 Table of contents. Introduction Literature review on the financial cycle What indicators are used to assess the financial cycle? The methodology Constructing a potential financial cycle measure for Albania Evaluation of the best indicators of the financial cycle Conclusions References

3 . Introduction The financial cycle plays an important role in increasing the sustainability of the financial system. Since the financial crisis, policymakers have been significantly reoriented toward macroprudential perspective, attempting to address the procyclicality of the financial system. The literature on the financial cycle measures is still young but it is closely tied to the rich literature of the booms and the busts of the financial system interlinked with the real sector. Following the financial crisis, macroprudential policies obtained an uprising significance. Subsequent to the assessment of systemic risk and the factors contributing to the crisis materialization, an important aspect is the construction of the macroprudential policy instruments, which are used to mitigate the fluctuations of the financial cycle in the economy. The fluctuations of the financial indicators or their impact on the financial system in terms of contractions or booms are accompanied with an increased exposure towards the financial crisis materialization. For this reason, estimating the financial cycle and the different phases of financial stress are crucial to the development of macroprudential policies and appropriate instruments that aim at mitigating the systemic risk. The literature in the aftermath of the financial crisis focuses on the preliminary assessment of the financial cycle. However, unlike the business cycle indicators, there is no single indicator defined that could capture the financial cycle and that could be used to quantify it. Overall, the financial cycle distinguishes from the business cycle due to its amplitude and its frequency (Stremmel, 205). The financial cycle develops usually in the medium-term compared to the short-term developments that characterize the business cycle. Moreover, the financial cycle is characterized by higher amplitudes and lower frequencies compared to the business cycle. Drehmann (202) finds an important relation between these low frequency financial cycles and the occurrence of financial crisis, suggesting that these cycles have an important role in explaining financial stress episodes. Borio (202) defines the financial cycle as an interaction between perceptions of value and risk, attitudes towards risk and financing constraints, which translate into booms followed by busts, binding this concept closely to that of the procyclicality of the financial system. These interactions could amplify economic imbalances accompanied by the lack of macroeconomic stability and/or could threaten financial stability. The aim of this paper is to estimate the financial cycle in Albania by creating a quantifying synthetic measure as part of the new macroprudential framework which is being developed by the Bank of Albania. This synthetic measure incorporates different elements from various segments of the financial system as well as their interlinkages into one single composite indicator. The paper is organized as follows. The second section reviews the existing literature on the financial cycle. The third section analyses the variables that could be considered to estimate the financial cycle in Albania. The fourth section explains the methodology implemented in constructing the financial cycle, meanwhile the fifth and the sixth section present the estimation results of the financial cycle in Albania. Finally, the last section concludes. 3

4 2. Literature review on the financial cycle The existing literature on the financial cycle is new and quite limited compared to the literature on the business cycle. The recent global financial crisis largely illustrated the real impact of systemic risk materialisation mainly related to a build-up of various macro-financial imbalances and vulnerabilities that can be associated to fluctuations in the financial cycle (Schuler et al., 205). As a result, the estimation of the financial cycle plays an important role in the framework of macroprudential policy and the activation of certain instruments before the systemic risk materializes. However, the literature has yet to identify and generally agree on a single definition of the financial cycle while this indicator remains significantly less covered compared to the business cycle indicator. The existing analyses on the financial cycle are quite limited and still inadequate at being easily applied in macroprudential framework. Even though there is no consensus in defining the financial cycle yet, according to Borio (202) the financial cycle can be determined as self-reinforcing interactions between perceptions of value and risk, attitudes towards risk and financing constraints, which translate into booms followed by busts, closely tying this concept to that of the procyclicality of the financial system. These interactions can amplify economic fluctuations and possibly lead to serious financial distress or macroeconomic imbalances. Drehmann et al. (200) propose that the financial cycle should be defined as the distance between two financial crisis, recommending the credit-to-gdp gap as a good starting point, due to its proper attributes to capture the performance of the crisis (Gonzalez et al., 205). Despite the lack of a clear definition for the financial cycle, the existing literature cites some distinctive features of the financial cycle. According to Drehmann et al (202), the financial cycle can be described mainly in terms of co-movements of credit and property prices. These indicators are closely connected to each other, especially in low frequencies, confirming the importance of credit in financing the construction sector and the purchases of real estate properties. The financial cycle shows lower frequencies than the traditional business cycle which has a length that varies between and 8 years, considering the cyclical component of the Gross Domestic Product. The financial cycle is longer than the business cycle (about four times) and it shows higher amplitudes. Drehmann et al. (202) identifies short-term cycles of -8 years like the business cycle, and medium-term cycles of 8-30 years. They identify the financial cycle as a medium-term phenomenon that tends to peak before a financial crisis, giving essential effects on macroeconomic indicators, and they construct it for an a priori longitude of 8 to 30 years (Strohsal et al, 205). Therefore, the financial cycle tends to extend in time, especially in the upswing phase. The contraction phase of the financial cycle also lasts for several years, unlike the recession of the business cycle which may not exceed the timeframe of one year (Schuler et al, 205). As a result, recessions that coincide with financial cycle contractions last longer than other recessions. Thus, the role of the financial cycle is very important in facilitating the expansion of the real economy, as well as promoting its further contraction (Einarsson et al 205). Moreover, the length and amplitude of the financial cycle depends on a country's policies regime (Borio, 202). Three factors are particularly important in this regard: the financial regime, the monetary regime and the regime of the real economy (Borio and Lowe (2002), Borio (2007)). The 4

5 financial development is reflected in the contraction of financial constraints and in the financial resources that affect the structure of the financial cycle. The inflation targeting monetary policy is active also during the peaks of the financial cycle, which are characterized by higher inflation rates, thus influencing both the length and extension of the financial cycle. Finally, the openness of the real economy to the financial markets increases the exposure to global financial crises and systemic risk, thus complicating the impact channels of the financial markets. Apart from changes in length, frequency and amplitude, the literature argues that the cyclical behavior of the financial indicators could be understood as a pure reflection of the real economy, but also as a result of major changes in the general perception and attitude towards macro-financial risks. In fact, the empirical findings suggest that the turning points of the financial cycle are generally associated with periods of financial stress, financial crisis or banking crisis. Largely, these periods of crisis that stem within a certain economy (thus, not originating from external exposure losses) occur at the peak of a financial cycle or very close to it. This connection helps explain an empirical "rule", whereby recessions coinciding with contraction periods of the financial cycle are especially more severe. Under these conditions, the estimation of this cycle helps to identify risks arising from financial stress in almost real time and with a high accuracy level. The construction of a financial cycle refers to the co-movement of a number of indicators in a given period of time. Combining different component indicators gives better signals of the financial cycle compared to a single indicator. The literature shows that the best representative of the cycle is given in terms of interlinkages between credit indicators and real estate indicators, even though the inclusion of other indicators could provide quite an important supplementary information. 3. What indicators are used to assess the financial cycle? As risk perceptions and attitudes towards the financial cycle cannot be measured directly, it is somewhat unclear which group or which combination of selected indicators can better reflect the financial cycle. The selection of the indicators is still preliminary and it is based on the mentioned literature, economic logic, as well as the time extent of the data series. To construct the financial cycle in Albania we relied on the alignment of Drehmann et al. (202) and Stremmel (205) by combining different time-series which are filtered to obtain the cyclical component. In the notion of adequate financial cycle determinants, we select a set of explanatory variables which are fundamental for their contribution in the assessment of systemic risk. The literature focuses mainly on various credit indicators and the price of non-financial assets. The credit cycle, reflected in the financing restrictions, changes of banks liquidity, financial leverage and other indicators based on the balance sheet of the banking sector, has a direct impact on the economic activity and, consequently, on the financial stability of a country. Consequently, the 5

6 literature on the financial cycle widely uses the credit indicators to assess the relationship between the financial system and the real economy (Einarsson et.al, 205). Generally, in accordance with the expectations on the financial cycle, developments in the credit cycle have a medium-term dimension, unlike business cycles that demonstrate short-term fluctuations (Aikman et.al, 205). Another important indicator of the interaction between financing constraints and the perception of value and risk is the real estate prices. Their role is generally in the center of the literature on financial peaks and contractions of a cycle, characterized by their growth right ahead of a crisis, followed by a significant and stable drop after the lowest point of contraction (Einarsson et. al, 205). The corresponding performance of credit indicators and asset prices provides reasonable information on the financial cycle development. According to Schuler et al (205), credit indicators and asset prices show similar cyclical frequencies and improve the predicting power of crises for longer time horizons. Overall, we have included seven indicators as potential components of the financial cycle based on the mentioned literature, four of which describe financial sector developments, while the other three belong to the banks balance sheet, focusing directly on the behavior of the banking sector. To capture the behavior of a financial cycle it is necessary to select data that have the longest possible time extent. However, for Albania, the timespan of the data is a limiting factor in the selection of indicators. The best we could do is to use the belowmentioned variables on quarterly basis with the largest possible timespan of 993Q4-205Q4. To assess the credit developments, we have included the following variables: ) credit-to-gdp ratio as an indicator of macroprudential policy analysis (Detken et. al 204). 2) annual credit growth rate (nominal), to assess the acceleration rate of credit expansion. The usage of growth rates aims at capturing the different fluctuations of the indicators that assess developments in credit and asset prices. To evaluate the performance of asset prices we have included the performance of real estate prices through: 3) House price index and 4) Annual growth of housing prices. The second group of indicators aims to assess the important role of the financial institutions balance sheet in the development of financial cycles. The information that comes from the banks balance sheet can shed light on the role of the banking sector to shocks arising from interactions within the financial system. During periods of financial cycle expansion, the financing standards are generally eased due to excess liquidity, allowing an expansion of banks' balance sheet and consequently of the other sectors of the economy, while this result is reversed during periods of contraction. Consequently, the use of banks' balance sheet information may also help in assessing their role in the amplification of shocks arising from macro-financial interactions (Einarsson et.al, 205). In this 6

7 regard, to assess the characteristics of the banking sector behavior, three indicators are included that refer to banks' balance sheet: 5) the ratio of short-term funding to total assets 6) net income to total assets 7) the ratio of total loans to total assets The first indicator "the ratio of short-term funding to total assets", aims to capture the cyclical behavior of banks funding which varies depending on the financial cycle developments. The indicator of net income to total assets" measures the profitability of the banking sector through the financial cycle. The ratio of "total loans to total assets" assesses the level of banks lending during the cycle measuring the expansion of credit activity in the balance sheet of the banking sector. All indicators are standardized to ensure the comparability of their units. 4. The methodology Synthetic indicators of the financial cycle are derived by aggregating various financial indicators of the cycle. A synthetic indicator implies that this composite indicator includes the impact that arises from different indicators without an analytical quantification of the relationship between them. The first step in this process is to derive the cyclical component for each of the indicators presented above. A frequency-based filter is a technique that studies the behavior of cyclical movements distinguishing the cyclical patterns of the time series. Two types of filters dominate in the contemporary literature to distinguish the cyclical behavior: the Hodrick-Prescott filter (HP) and the band-pass filter (BP). The HP filter separates the cyclical component and the historical trend (trend) from data series by applying a conditional function that penalizes deviations from historical trend by using predefined weights. The BP filter is basically a two-sided moving average that filters out certain frequencies in the time series. This filter is widely used in the literature of the financial cycle as an appropriate tool in calculating the cyclical components of the individual series. Using this methodology enables smoother cycles than the series filtered through the HP filter and facilitates the comparability between the time series. More specifically, in each period t =,... N, the performance of each indicator includes two components: the long-term trend or trend (L t ) and the cyclical behavior or deviation cycle (D t ) as follows: T t = L t + D t The long-term component L t and the cyclical deviation component D t cannot be monitored so they have to be assessed: T t = L t +D t Based on Christiano and Fitzgerald (2003). 7

8 The literature suggests two methods on assessing the long-term components: the band-pass filter Baxter and King (999) otherwise known as the BK filter and Christiano and Fitzgerald (2003), otherwise known as the CF filter. In the case of the BK filter, it is necessary to determine the specific lag length used in the assessment of the permanent components and the symmetry (equal distribution of weights for the leading and lagging indicators), while the CF filter enables the data itself to dictate the weights (Boshoff, 200). In case of high frequencies, both filters provide the same results. However, in case of low frequencies commonly used for the financial cycle, the empirical findings suggest that the CF filter evaluates more accurately the long-term component of the indicator (Zarnowitz and Ozyildirim, 2006). In this paper, the band-pass filter is implemented through the CF approach by selecting the specification that makes no correction to the unit roots and does not eliminate the trend (historical trend) of the selected indicators, to keep the distortive filter effects to a minimum level. Given that the literature argues that the length of the financial cycle is 4 times the length of the business cycle (Ravn and Uhlig, 2002), we assume that the duration of a financial cycle extends over a period of 32 to 20 (ie, medium-term cycle of 8-30 years). The construction of a synthetic indicator as a possible measure of the financial cycle involves two steps: a) The application of a frequency-based filter on (BP) for each of the selected time series, to compare the behavior of medium-term cycles. b) The combination of several financial indicators filtered for the construction of a single composite indicator, which will serve as a measure of the financial cycle. In the first step we construct the cyclical medium-term components for each of the indicators: credit-to-gdp ratio, the house price index, credit growth, the housing price growth, the bank funding ratio, the net income to total assets ratio and the total loans to total assets ratio. The graph of cyclical movements for each financial indicator shows the cyclical characteristics of indicators and their potential usage. From the charts below we can detect a limited number of full cycles during the timespan taken in consideration. Graph shows that the cyclical component of credit and house prices co-move together. The booms and contractions occur within the same timeframe, with similar frequencies, although with different amplitudes. For both indicators, their growth rates tend to culminate faster that the performance of the stock indicators, while the credit-to-gdp ratio and the house price index tend to adjust more gradually. Observing the financial stress periods in Albania 2, the annual credit growth and the growth rate of the house price index behave as leading indicators, while the other variables are quite lagged. 2 As identified by Minka and Kota (207) and Kota and Saqe (203). 8

9 993T4 994T3 995T2 996T 996T4 997T3 998T2 999T 999T4 2000T3 200T2 2002T 2002T4 2003T3 2004T2 2005T 2005T4 2006T3 2007T2 2008T 2008T4 2009T3 200T2 20T 20T4 202T3 203T2 204T 204T4 205T3 993T4 994T3 995T2 996T 996T4 997T3 998T2 999T 999T4 2000T3 200T2 2002T 2002T4 2003T3 2004T2 2005T 2005T4 2006T3 2007T2 2008T 2008T4 2009T3 200T2 20T 20T4 202T3 203T2 204T 204T4 205T3 Graph. Credit and asset prices indicators Cyclical movements of the credit and asset price indicators Crisis HPI Credit-to-GDP annual credit growth annual HPI growth Source: Bank of Albania, authors calculations The filtered indicators based on the balance sheet data (graph 2) vary in frequency and amplitude compared to the credit and asset prices (housing) indicators. However, the net-income-to-totalassets ratio and and the short-term funding to total assets ratio behave as leading indicators of the financial stress periods, while the total-loans-to-total-assets ratio culminates at the end of the identified financial stress period. Graph 2. Balance-sheet data indicators Cyclical movements of balance sheet data indicators Crisis Short-term funding to total assets Net income to total assets Loans to total assets Source: Bank of Albania, authors calculations 9

10 These results show the relative importance that each of these indicators might have in constructing the financial cycle. However, the validity of these cyclical movements is limited as long as an indicator may not capture certain developments in different financial markets. For this reason, cyclical measures must be built for the entire financial system, which should be validated as appropriate before drawing any conclusions. 5. Constructing a potential financial cycle measure for Albania In order to calculate a synthetic indicator of the financial cycle, the indicators presented above are combined in some aggregated indicators following the approach of Stremmel (205), as presented in Table : Table. Different potential financial cycle measures Financial cycle measure FC FC2 FC3 FC4 FC5 FC6 FC7 Indicators Credit-to-GDP ratio Credit-to-GDP ratio, House price Index Credit-to-GDP ratio, House price Index, Credit growth Credit-to-GDP ratio, House price Index, Credit growth, House price growth Credit-to-GDP ratio, House price Index, Credit growth, Short-term funding to total assets Credit-to-GDP ratio, House price Index, Credit growth, Net income to total assets Credit-to-GDP ratio, House price Index, Credit growth, Total loans to total assets The first synthetic indicator FC is a measure based on only one indicator. This filtered indicator based on the credit-to-gdp ratio, well-known in the macroprudential literature, is a reliable early warning indicator of financial crises and its explanatory power can gradually increase with the involvement of other indicators. For this reason, the credit-to-gdp ratio is used as a starting point in calculating a composite indicator to measure the financial cycle, by further adding other selected indicators. The length of the time series used is an important issue in the construction of these financial cycle measures. The periods which include all the component indicators are shorter for the potential measures between FC2 to FC7, due to the lack of data on housing prices before the second quarter of 998. The potential measures (graph 3) show similar patterns, although they share different amplitudes and different turning points in time. The graphical illustration of various cycle measures do not provide a final answer, but they do give some intuition on selecting the best potential measure of the financial cycle. For this reason, we also rely on statistical methods. 0

11 993Q4 994Q3 995Q2 996Q 996Q4 997Q3 998Q2 999Q 999Q4 2000Q3 200Q2 2002Q 2002Q4 2003Q3 2004Q2 2005Q 2005Q4 2006Q3 2007Q2 2008Q 2008Q4 2009Q3 200Q2 20Q 20Q4 202Q3 203Q2 204Q 204Q4 205Q3 Graph 3. Potential financial cycle measures Potential financial cycle measures Financial stress FC FC2 FC3 FC4 FC5 FC6 FC Source: Authors calculations 6. Evaluation of the best indicators of the financial cycle To determine the quality of the potential financial cycle measures, we evaluate the relationship between different synthetic indicators and financial crisis periods, focusing on how they coincide in time, but without aiming at using the financial cycles as an early warning indicator of the financial crisis. In line with the macroprudential literature, we estimate a logit model for each of the indicators of the financial cycle: Crisis i,t = C + CF i,t The dependent variable in this case is the "crisis" (taking the value of in periods of financial stress as identified above and 0 otherwise) and the independent variables are the constant C and each potential financial cycle measure. The evaluation procedure determines the potential of each cycle measure in explaining the periods of financial crisis or financial stress. In addition, to evaluate the

12 suitability of each potential measure of the financial cycle, we use the statistical method of the AUROC curve, based on Drehmann and Juselius (204). The signaling approach is a non-parametric approach, which is used to rank the indicators based on their ability to provide good signals for future crises or to avoid false alarms. More specifically, each financial cycle measure could provide four possible types of signals for the materialization of a financial crisis: Table 2. Evaluation of the signaling ability of the financial cycle Crisis during Q No crisis during Q Signal A C No signal B D Where:. The rate of true positive signals is measured as A/(A+C) 2. The rate of false alarms is measured as B/(B+D) 3. The noise to signals ratio is measured as {B/(B +D)}/{A/(A+C)} for each quarter Q from 2 to 0 before the crisis or financial stress periods The signaling ability of the financial cycle indicators is measured by plotting the AUROC curve (area under the receiver operator characteristic curve) which evaluates the accuracy of forecasting the financial stress periods (crisis) for each critical level of the financial cycle indicators. The noise ratio (the ratio of false alarms where the financial cycle measure reports a crisis but in reality there is no crisis) is placed against the true signal ratio (the ratio of accurate signals, where the financial cycle measure reports the crisis and there is crisis indeed) for each possible value of the financial cycle (Valinskyte and Rupeika, 205). The comparison of the accuracy of each indicator is quantified by measuring the area under the AUROC curve. A value of 0.5 indicates that the financial cycle measure does not provide enough information to forecast the financial crisis. The value indicates that the financial cycle measure gives perfect information for predicting the financial crisis, while values between serve to measure the indicator that has the best performance 3. As a perfect indicator does not exist, there is always a "trade-off" between missed crises and false alarms. 3 A general classification on the forecasting accuracy for the AUROC indicator is: a = excellent b = good c = satisfactory d = poor e = fail 2

13 Graph 4. An example of the AUROC curve Graph 5 shows the AUROC for each of the financial cycle measures, starting from 2 before the crisis up to the assessment of the crisis period itself. In total, 9 AUROC curves estimations were made, while in the graph the value of the surface for each curve is shown. According to the results presented in the graph, the financial cycle measures are quite informative in regard to periods of crisis and financial stress, having obtained values above 0.5. For all the financial cycle measures, the AUROC fluctuates between the range of , with the highest level for FC3 and FC7. Graph 5. AUROC of financial cycle measures AUROC of financial cycle measures financial stress FC FC2 FC3 FC4 FC5 FC6 FC7 Source: Authors calculations Except the AUROC indicator, to summarize which of the financial cycle measures FC-FC7 estimates the financial cycle more accurately, expert judgment is also required regarding the peaks and 3

14 993Q4 994Q3 995Q2 996Q 996Q4 997Q3 998Q2 999Q 999Q4 2000Q3 200Q2 2002Q 2002Q4 2003Q3 2004Q2 2005Q 2005Q4 2006Q3 2007Q2 2008Q 2008Q4 2009Q3 200Q2 20Q 20Q4 202Q3 203Q2 204Q 204Q4 205Q3 troughs of the financial cycle assessed by each measure. Starting from the first measure (FC), which summarizes only the developments of credit to GDP ratio, this measure does not show a satisfactory performance in assessing the financial cycle. A similar assessment is given for FC2 also, supported by the low levels of the AUROC. On the other hand, FC3 and FC7 measures summarize developments not only in stock level but also in terms of flows of the selected indicators, assessing the financial cycle more accurately. These estimations are also empirically based by the high level of the AUROC for the period of 8 to 4 before the crisis, a period deemed as most appropriate for measuring the financial cycle accuracy. In addition, the choice between FC3 and FC7 measures also requires expert judgment considering: a. which of the two measures better assesses the boom before periods of financial stress (FC3 considers the peak at mid-2008, while FC7 considers the peak at mid-2009) and b. which of the measures can be interpreted more easily in the financial cycle terms (FC3 enables an easier interpretation in terms of the credit-to-gdp ratio, the house price index and credit growth rate while interpreting the credit-to-total-assets ratio (FC7) is more difficult). Taking into account the graphical illustration and the statistical evidence, as well as the data availability, we conclude that the indicators of credit-to-gdp ratio, the house price index and the credit growth rate (FC3) are the best choice in constructing a synthetic indicator of measuring the financial cycle. In addition, Graph 6 presents a comparison between the performance of the financial cycle and the assessment of the business cycle 4. The performance of both indices for Albania is in accordance with the literature findings of other countries. Graph 6. The financial cycle and business cycle in Albania financial stress Financial cycle, FC3 Business cycle, rhs Source: Authors calculations and Monetary Policy Department 4 Estimated by Monetary Policy Department, Bank of Albania. 4

15 In general, the average duration of the financial cycle is greater than that of the business cycle. Also, the business cycle has a higher variance and is more volatile than the financial cycle. After the first period of the financial crisis (late 2008), it appears that the slowdown of the financial cycle in Albania has had a strong impact on the country's economic developments, reflected in a rapid contraction of the business cycle. According to Stremmel (205) these findings have a significant impact on building appropriate macroprudential policies. In general, the booms of the financial cycle are more associated with the financial crisis than with the business cycle. Also, these booms are followed by a rapid contraction of the economic activity (Drehman et.al, 202), stressing the importance of their timely estimation once again. 7. Conclusions This paper identifies key components of the financial cycle in Albania, its construction techniques and differences between potential indicators that could serve as measures of financial cycle. Based on statistical and graphical evaluations, the results show that indicators of credit-to-gdp ratio, the house price index and the credit growth rate are the best choices to construct a synthetic index to measure the financial cycle. The results also show that the financial cycle is positively correlated with periods of financial stress, which should be considered appropriately in policies that could be implemented. The financial cycle could be used in the context of assessing the cyclical position of the financial system and also to provide the necessary signals where macro-financial vulnerabilities are on the rise. The composite measure of the financial cycle can serve to different purposes in macro-prudential policies. From the perspective of financial stability, the state of the financial cycle and of its composing indicators is essential in determining the appropriate actions. The time of implementation of the macro-prudential measures in the financial cycle is crucial in minimizing economic costs. 5

16 8. References Borio (202), The financial cycle and macroeconomics: What have we learnt? Bank for International Settlements Boshoff, W.H. ( 200) Band-pass filters and business cycle analysis: high-frequency and mediumterm deviation cycles in South Africa and what they measure, ERSA Working Paper 200. Detken et al. (204), Operationalising the countercyclical capital buffer: indicator selection, threshold identification and calibration options Drehmann M., Borio C., Tsatsaronis K., (20): Anchoring countercyclical capital buffers: the role of credit aggregates, BIS Working Papers Drehmann M., Borio C., Tsatsaronis K., (202): Characterising the financial cycle: don t lose sight of the medium term!, BIS Working Papers No 380 Einarsson B., Gunnlaugsson K., Ólafsson T., Pétursson T., (205): The long history of financial boombust cycles in Iceland Gonzalez R., Lima J., Marinho L., (205): Business and Financial Cycles: an estimation of cycles length focusing on Macroprudential Policy Kota V., Saqe A. (203): A financial systemic stress index for Albania Ravn. M., Uhlig H., (2002): Notes on adjusting the Hodrick-Prescott filter for the frequency of observations, The Review of Economics and Statistics Schuler Y., Hiebert P., Peltonen T., (205): Characterising the financial cycle: a multivariate and time-varying approach, ECB working paper Stremmel H. (205): Capturing the financial cycle in Europe, Working Paper Series (ECB) Strohsal T., Proańo C., Wolters J., (205): Characterizing the financial cycle: evidence from a frequency domain analysis Valinskytė N., Rupeika G., (205): Leading indicators for the countercyclical capital buffer in Lithuania Minka O. dhe Kota V. (206), Early Warning Indicators of Financial Crisis, forthcoming Zarnowitz, V. and A. Ozyildirim (2006). "Time Series Decomposition and Measurement of Business Cycles, Trends and Growth Cycles." Journal of Monetary Economics53 (7). 6