Conditional Ideologically Driven Monetary Policy Cycles: Myth or Reality?

(1)

Rijksuniversiteit Groningen

Conditional Ideologically Driven Monetary Policy Cycles:

Myth or Reality?

Federico M. Giesenow†

Abstract

By employing newly constructed data sets, this paper examines the extent to which governments’ ideology and the degree of central banks’ autonomy jointly affect monetary policy. Throughout this document, the hypothesis that interest rates are lower during left-wing governments only if the monetary authority is not independent will be challenged. To achieve this goal, the robustness of this proposition will be tested under two alternative theoretical frameworks by means of a dynamic fixed-effects panel data approach. After incorporating a forward-looking central bank that works with volatile forecasts and after highlighting the role of the different monetary regimes, the preliminary results cast doubts on the strength of the aforementioned hypothesis.

Keywords: Monetary Policy, Ideology, Partisan Theory, Central Bank, Political Economy. JEL Classification: E52 – E58 – D72 – C23

† _{The author is a Research Master student at the University of Groningen. Email address:}_{f.m.giesenow@student.rug.nl}_.

(2)

1. Introduction and Theoretical Context

In 2008 the world economy was shocked by a gruesome financial crisis from which still today many countries have not been able to fully recover. The presence back then of a massive financial bubble is a fact that not many scholars will currently disagree on. However, reaching a consensus on the economic or institutional mechanisms that generated and sustained that bubble has proven to be a herculean task. In this context, some economists believe that the United States (U.S.) Federal Reserve and other leading monetary authorities in the world could claim a substantial share of responsibility (Roubini, 2005). If this were the case, it would indicate that little has been learnt from the painful experiences of previous crises. Only by fully understanding the process through which central banks’ decisions are made, it will be possible to prevent another financial meltdown. Not surprisingly, since 2008 the study of the determinants of central banks’ decisions has regained momentum in the field of Political Economics. In particular, the focus of this document will therefore be the influence of governments’ ideological agenda on the actions of the monetary authority.

The influence of political ideology on the state of the economy is not per se a new concept; in the past, Partisan Theories (Hibbs, 1977; Alesina, 1987) were built upon the idea that political parties endorse economic policies that are aligned with the preferences of their voters. Hibbs’ traditional partisan theory presented the notion that left-wing governments will tend to favour pro-growth policies over the control of inflation, whereas right-wing governments will do the opposite (Hibbs, 1977). Alesina’s rational approach to partisan theory foresees that, within a rational expectations framework with optimizing subjects and in the context of election uncertainty, right-wing governments will tend to show higher post-election levels of unemployment (Alesina, 1987; Alesina & Sachs, 1988). From the particular point of view of the monetary policy, the partisan theories would imply lower interest rates under leftist administrations rather than under right-wing ones. This theorized behaviour of the interest rate will be referred as ideologically driven monetary policy cycle (IDMPC) throughout this document.

(3)

of a more relaxed (pro-growth) monetary policy. However, Sakamoto’s research is based on yearly figures, forcing him to start his sample period as early as in the 1960’s in order to gather enough critical mass of data. In this context, Belke & Potrafke (2012) addressed the aforementioned empirical contradictions and also improved previous findings by conducting a fixed-effects panel data analysis. Based on a backward-looking central bank, their research supports the idea of a conditional IDMPC, in which short-term interest rates will be lower during left-wing governments only when the monetary authority can be significantly influenced by the government (with the latter effect being approached by a comprehensive measure of Central Bank Independence, or CBI for short). Within the same framework, independent central banks will have the opposite effect, causing interest rates to rise.

(4)

The second theoretical aspect to be addressed is related to the heterogeneity present in the definition of Central Bank Independence. Institutional aspects, such as laws and regulations, constitute the de jure part of the CBI definition, which only describes to what extent the monetary authority is independent from a legal point of view. At the same time, the real or practical extent to which this autonomy is actually exercised constitutes the de facto part of the concept (Cukierman, 2007). Although many attempts have been made to obtain more comprehensive measures of CBI (Alesina & Summers, 1993; Klomp & de Haan, 2009), still no single index is able to fully capture the vast complexity embedded in the difference between these two concepts. Consequently, it is important to account for variables that may create a wedge between both de jure and de facto concepts of autonomy, such as a country’s type of monetary regime (Arnone, Laurens, Sommer, & Segalotto, 2007). A monetary authority which is porous to political pressure, but at the same time is confined by other institutional arrangements to follow a given policy objective, will find itself limited in the extent to which it can accomplish the government’s goals. A good example of this idea could be Inflation Targeting (IT), which is a monetary regime that embodies a strong institutional commitment that forces central banks to keep inflation within a certain range (Amato & Gerlach, 2002; Roger & Stone, 2005; Rose, 2007). With this in mind, and taking advantage of an innovative and recently published classification of countries’ monetary policy strategies (Samarina, 2014), it will be possible to test the wide-ranging validity of the conditional IDMPC hypothesis across particular monetary regimes.

Finally, and as a last test of the resilience of the proposed conditional ideologically driven monetary policy cycle, this hypothesis will be the object of a sensitivity analysis. The aim of this examination is to measure the impact of excluding specific countries from the sample. An important implication of the results in previous research that have led to the conditional IDMPC hypothesis is that the magnitude of this effect should be the same for all countries included in the sample. If this holds, the results should be robust to the exclusion of single countries from the analysis.

The reminder of the document is structured as follows. Section 2 presents a description of the empirical model, the proposed innovations, and the main characteristics of the data used. Section 3 discusses the empirical findings, and Section 4 offers the conclusions and closing remarks.

(5)

2.1 Baseline Model

The baseline model consists of the model proposed in Belke & Potrafke’s (2012) paper. Therefore, the subjects of the study will be 23 countries from the Organisation for Economic Co-operation and Development (OECD) included in these authors’ research and the period of time considered will range from 1980 to 2005. Additionally, central banks’ monetary rule will be represented by a classic Taylor-rule (Taylor, 1993), which explains the behaviour of short-term nominal interest rates by accounting for the output gap and the rate of consumer inflation. The model also includes a one-period lag of the short-term interest rate as explanatory variable to reflect the interest rate smoothing behaviour of central banks (English, Nelson, & Sack, 2002). In addition to the main components of the traditional Taylor-rule, the model includes variables that represent governments’ ideology, Central Bank Independence, and the interaction term between these two. This interaction term will allow the examination of the influence of government ideology conditional on the different levels of central bank autonomy (CBI). The estimation method is a fixed-effect panel data model, which presents the following characteristics: 𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑅𝑎𝑡𝑒_𝑖𝑡 = 𝛽₁𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑅𝑎𝑡𝑒_𝑖𝑡−1+ 𝛽₂𝐼𝑑𝑒𝑜𝑙𝑜𝑔𝑦_𝑖𝑡+ 𝛽₃( 1 𝐶𝐵𝐼)𝑖𝑡+ 𝛽4𝐼𝑛𝑓𝑙𝑎𝑡𝑖𝑜𝑛𝑖𝑡+ 𝛽₅𝐺𝐷𝑃𝑔𝑎𝑝_𝑖𝑡+ 𝛽₆(𝐼𝑑𝑒𝑜𝑙𝑜𝑔𝑦 ∗ ( 1 𝐶𝐵𝐼))𝑖𝑡+ 𝜂𝑖+ 𝜇𝑡+ 𝜀𝑖𝑡 (1)

(6)

ideology, and the interaction term, then these variables will be normalized to allow direct interpretation of its coefficients. With these changes in mind, the baseline model looks as follows:

𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑅𝑎𝑡𝑒_𝑖𝑡 = 𝛽₁𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑅𝑎𝑡𝑒_𝑖𝑡−1+ 𝛽₂𝐼𝑑𝑒𝑜𝑙𝑜𝑔𝑦_𝑖𝑡+ 𝛽₃𝐶𝐵𝐷_𝑖𝑡+ 𝛽₄𝐼𝑛𝑓𝑙𝑎𝑡𝑖𝑜𝑛_𝑖𝑡+

𝛽₅𝐺𝐷𝑃𝑔𝑎𝑝_𝑖𝑡+ 𝛽₆(𝐼𝑑𝑒𝑜𝑙𝑜𝑔𝑦 ∗ 𝐶𝐵𝐷)_𝑖𝑡+ 𝜂_𝑖+ 𝜇_𝑡+ 𝜀_𝑖𝑡 (2) In this context, traditional partisan theories will predict a negative marginal effect of ideology. Additionally, a traditional Taylor-rule will also predict a positive sign for β4 and for β5.

Following the literature in this field, the effect of the interest rate smoothing parameter (β1)

should be positive and smaller than one. Although the sign and size of both β3 and β6 are not

clearly defined in the literature, the conditional IDMPC hypothesis will imply that the marginal effect of ideology at the maximum level of CBD should be negative and statistically significant. On the other hand, the marginal effect of ideology should be positive at the minimum of CBD. Incorporating a lagged dependent variable as regressor into the panel data analysis turns it into a dynamic one, in which the traditional fixed-effect estimators are biased according to Nickell (1981). In this case, and due to the small number of countries (N=23), the methodology proposed by Arellano & Bond (1991) to correct this particular problem will also be biased.1

Therefore, Bruno’s (2005) bias-corrected least-squares dummy variable estimation (LSDVC) routine for dynamic panels is applied.2_{This technique will not only control for Nickell’s bias,}

but also will become handy to assist in dealing with other problems faced throughout this analysis: missing information (unbalanced panel) and first-order serial correlation.3_{Both the}

backward-looking panel used to obtain the baseline approach as well as the newly constructed forward-looking panel suffer from the missing information problem. To deal with the potential bias created by the unbalanced panel, Bruno’s procedure implements a strictly exogenous selection rule to extend Bun & Kiviet’s (2003) bias approximation method. Additionally, Wooldridge’s test for serial correlation in the idiosyncratic errors shows the presence of this particular problem as well (Wooldridge, 2001).4_{Fortunately, the LSDVC approach can also}

cope with it under additional assumptions.5

1_{A drawback of Instrumental Variables (IV) and Generalized Method of Moments (GMM) estimators to cope with this problem is}

the fact that their properties only hold for large number of cross-sectional units (large N).

2_{Stata command xtlsdvc is used, with option “bb” for the initial values. Furthermore, 25 iterations are considered when}

bootstrapping the estimated standard errors. This procedure assumes that the independent variables are exogenous.

3_{An additional problem could be the non-stationarity of the time series considered. However, in this case the analysis follows}

Rudebusch (1993) and makes no claim that the short-term interest rate, inflation, and output gap series are non-stationary.

4_{The results of this test are included in the Appendix.}

5_{Bruno’s STATA routine requests starting values to correct for the biases. In this sense, the baseline model selects the Blundell &}

(7)

2.2 Innovations

The first innovation will be to consider the impact of a forward-looking central bank. In order to do so, the variables Inflation and GDPgap will be replaced by their expectational counterparts. This type of data comes from two different sources, OECD Economic Outlook database and Consensus Economics Inc., each with its particular characteristics. The former, presents data that allows the calculation of semi-annual figures for different planning periods. The latter offers only 1 planning period (12 months ahead), but provides monthly figures and estimates of the intra-temporal volatility of the forecasts.

Using data from the OECD Economic Outlook publications, inflation and GDPgap will be replaced in model (2) by private expectations about the future behaviour of these variables. The initial time horizon to be considered for the forward-looking variables will be the 1-year period as this is the time frame most commonly found in the literature. Additionally, and to cope with the possibility of central banks working with other time horizons, both a 6-month and an 18-month frames will also be used. It may seem odd to use the 18-months planning horizon in this context, but in the past banks such as the New Zealand’s Reserve Bank have set their monetary policy response horizon at 6 to 8 quarters ahead (Plantier, 2002). In a departure from the baseline model, and forced by the lack of quarterly forward-looking figures from this particular source, this part of the analysis will consider only semi-annual data. Further details on the characteristics of the new variables will be presented in the next subsection. To summarize, once the forward-looking variables have been incorporated into the analysis, the resulting alternative model will be (for the 1-year case):

𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑅𝑎𝑡𝑒𝑖𝑡 = 𝛽1𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑅𝑎𝑡𝑒𝑖𝑡−1+ 𝛽2𝐼𝑑𝑒𝑜𝑙𝑜𝑔𝑦𝑖𝑡+ 𝛽3𝐶𝐵𝐷𝑖𝑡+ 𝛽4 𝐸(𝐼𝑛𝑓𝑙𝑎𝑡𝑖𝑜𝑛)𝑖𝑡 +

𝛽5 𝐸(𝐺𝐷𝑃𝑔𝑎𝑝)𝑖𝑡 + 𝛽6(𝐼𝑑𝑒𝑜𝑙𝑜𝑔𝑦 ∗ 𝐶𝐵𝐷)𝑖𝑡+ 𝜂𝑖+ 𝜇𝑡+ 𝜀𝑖𝑡 (3)

where the E(.) operator indicates that the variable is an expectation.

(8)

uncertainty surrounding the forecasting profession. 6_{In the past, most of the empirical}

research in this field has overlooked this particular variable focusing more on the ability of central banks to decrease the –average- forecast errors or to lower the volatility of target variables such as growth or inflation.7_{In this case, the variable of interest is the distribution of}

the forecasts at a given moment of time, which is assumed to be predetermined. Additionally, it is not unrealistic to assume that the central bank knows it at the moment of deciding the level of the interest rate. In this context, two alternative models will be implemented. The first one will be just a recalculation of model (3) but using Consensus Economics’ data. The second one will be a departure from the original Taylor-rule that incorporates the variability of the expected values of inflation (Vol.infl) and GDPgap (Vol.gdp) as moderators. The expanded model will then look like this:

𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑅𝑎𝑡𝑒𝑖𝑡 = 𝛽1𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑅𝑎𝑡𝑒𝑖𝑡−1+ 𝛽2𝐼𝑑𝑒𝑜𝑙𝑜𝑔𝑦𝑖𝑡+ 𝛽3𝐶𝐵𝐷𝑖𝑡+ 𝛽4𝐸(𝐼𝑛𝑓𝑙𝑎𝑡𝑖𝑜𝑛)𝑖𝑡+ 𝛽₅𝐸(𝐺𝐷𝑃𝑔𝑎𝑝)_𝑖𝑡+ 𝛽₆(𝐼𝑑𝑒𝑜𝑙𝑜𝑔𝑦 ∗ 𝐶𝐵𝐷)_𝑖𝑡+ 𝜑_𝑖𝑡+ 𝜂_𝑖+ 𝜇_𝑡+ 𝜀_𝑖𝑡 (4) Where, 𝜑𝑖𝑡 = 𝛽7(𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑅𝑎𝑡𝑒𝑖𝑡−1∗ 𝑉𝑜𝑙. 𝐼𝑛𝑓𝑙𝑖𝑡) + 𝛽8(𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑅𝑎𝑡𝑒𝑖𝑡−1∗ 𝑉𝑜𝑙. 𝑔𝑑𝑝𝑖𝑡) + 𝛽9(𝐼𝑑𝑒𝑜𝑙𝑜𝑔𝑦 ∗ 𝑉𝑜𝑙. 𝐼𝑛𝑓𝑙)𝑖𝑡+ 𝛽10(𝐼𝑑𝑒𝑜𝑙𝑜𝑔𝑦 ∗ 𝑉𝑜𝑙. 𝑔𝑑𝑝)𝑖𝑡 + 𝛽11(𝐶𝐵𝐷 ∗ 𝑉𝑜𝑙. 𝐼𝑛𝑓𝑙)𝑖𝑡+ 𝛽12(𝐶𝐵𝐷 ∗ 𝑉𝑜𝑙. 𝑔𝑑𝑝)𝑖𝑡 + 𝛽13[𝐸(𝐼𝑛𝑓𝑙𝑎𝑡𝑖𝑜𝑛) ∗ 𝑉𝑜𝑙. 𝐼𝑛𝑓𝑙]𝑖𝑡+ 𝛽14[𝐸(𝐺𝐷𝑃𝑔𝑎𝑝) ∗ 𝑉𝑜𝑙. 𝑔𝑑𝑝]𝑖𝑡 + 𝛽15(𝐼𝑑𝑒𝑜𝑙𝑜𝑔𝑦 ∗ 𝐶𝐵𝐷 ∗ 𝑉𝑜𝑙. 𝐼𝑛𝑓𝑙)𝑖𝑡+ 𝛽16(𝐼𝑑𝑒𝑜𝑙𝑜𝑔𝑦 ∗ 𝐶𝐵𝐷 ∗ 𝑉𝑜𝑙. 𝑔𝑑𝑝)𝑖𝑡 + 𝛽17𝑉𝑜𝑙. 𝑖𝑛𝑓𝑙 + 𝛽18𝑉𝑜𝑙. 𝑔𝑑𝑝

Finally, the last innovation takes advantage of Samarina’s (2014) classification of monetary strategies. The idea at this stage simply consists in splitting the same sample used in the baseline model into smaller subgroups using the monetary policy strategies followed by a given country at a given time as separation criteria. In this case, a backward-looking central bank is used again in the analysis. Among other objectives, countries in the sample could target inflation, exchange rates, money growth or have no particular target defined at all. Consequently, the same econometric analysis conducted in the baseline case (2) will be performed on those groups for which the real -de facto- extent of the central bank

6 _{The relative standard deviation, also known as coefficient of variation, is the standard deviation scaled by the absolute value of}

the mean.

7 _{For a review of the empirical evidence on the variability of professional forecasts and the central bank activities, see Middeldorp}

(9)

independence is identified. In particular, countries following an Inflation Targeting regime are assumed to have restricted the influence of government’s ideology to a minimum. Therefore, and even if the inferences from Arnone, Laurens, Sommer, & Segalotto, (2007) are correct, the conditional ideologically driven monetary policy cycle (IDMPC) should be reflected by a positive marginal effect of ideology on the interest rate at average levels of CDB. The same effect is assumed to be present when dealing with those countries following an Exchange Rate Targeting (EXRT) regime. Under high mobility of capital, EXRT regimes impose severe restrictions to the discretionary movement of the interest rates.

2.3 Sensitivity analysis

(10)

Table 1: Descriptive Statistics. Short-term Nominal Interest Rate. a_{Country List.}

Country Mean St. Dev. N Min Max

Australia 9.6 4.6 104 4.3 19.5 Austria 6.6 2.5 76 3.3 12.4 Belgium 8.6 3.6 76 3.0 16.2 Canada 7.7 4.2 104 2.1 20.7 Switzerland 4.0 2.6 104 0.3 10.1 Germany 5.8 2.5 32 3.2 9.8 Denmark 8.1 4.6 104 2.1 19.9 Spain 12.5 4.2 76 3.7 22.7 Finland 10.3 4.2 76 3.1 17.5 France 9.0 3.4 76 3.3 17.4 United Kingdom 8.8 3.8 104 3.5 17.7 Greece 17.0 5.0 84 3.5 24.0 Ireland 9.4 3.6 60 3.6 24.0 Iceland 11.3 7.6 72 4.5 37.7 Italy 12.8 4.2 76 4.0 20.5 Japan 3.6 3.2 104 0.0 12.6 Luxembourg 8.6 3.6 76 3.0 16.2 Netherlands 6.6 2.5 76 2.8 13.1 Norway 9.1 4.2 104 2.0 16.5 New Zealand 11.0 5.4 104 4.5 25.8 Portugal 14.8 5.6 76 3.7 24.9 Sweden 7.9 4.1 96 1.5 15.1 United States 6.8 3.9 104 1.1 18.4 Average/Total 9.01 5.21 1964 2.87 18.8

Notes: aAnnual percentage rate.

2.4. Data

Historical data for the variables included in the baseline approach can be obtained from the OECD’s online databases.8_{The GDP gap variable is not directly observable, but it is based on}

the published data of real GDP and obtained by applying the Hodrick-Prescott filter method (Giorno, Richardson, Roseveare, & van den Noord, 1995).9_{Additionally, the baseline model}

includes an index of government ideology proposed by Potrafke (2009). This index categorizes countries’ governments on a left-to-right scale that ranges from 1, indicating the presence of a strong right-wing ideology, to a value of 5, showing a strong left-wing administration. If the country’s government consists of a centrist party, or a balanced coalition of both right and left-wing parties, the index will take a value of 3. For those periods in which governments from different ideology are alternating in office, the index has a value according to the ideology of the party that spent more days in office. Another important variable used in the baseline model

8_{Available at}_{http://stats.oecd.org}_{. The complete dataset for the replication of the baseline model was generously provided by}

Prof. Dr. Niklas Potrafke, to whom the author of this document is grateful.

(11)

is an index of Central Bank Independence (CBI). This variable is constructed using the index provided by Arnone, Laurens, Sommer, & Segalotto, (2007), which in turn has been expanded by including additional periods by Klomp & de Haan (2009). Moreover, this particular measure of CBI is based on the methodology proposed by Grilli, Masciandaro, and Tabellini (1991), and therefore distinguishes between political autonomy (how objectives are determined) and economic autonomy (how instruments are determined). As stated in section 2, a measure of Central Bank Dependence (CBD) is constructed from the CBI variable described above. Central Bank Dependence can range from a maximum value of 1 to 0 at its minimum.

Following the baseline model, the countries under study are the following: Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Italy, Japan, Luxembourg, The Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, United Kingdom, and the United States. The empirical analysis in this paper uses data ranging from the first quarter of 1980 to the fourth quarter of 2005. However, for some countries, particular events limit the use of all the information available: in Germany’s case, the data considered only covers the period after the German Reunification (from first quarter 1991), Ireland only presents data since first quarter of 1984, Iceland since the first quarter of 1988, and Sweden’s information is only available starting in 1982. Additionally, for countries joining the European Economic and Monetary Union (EMU), information is restricted only to those periods in which their monetary policy was independent.10

Regarding the forward-looking approach, it was previously stated that this type of information is commonly not readily available; a feature that tends to limit this type of analysis. The solution chosen to overcome this problem has been to invest a considerable amount of time in manually processing and collecting forward-looking information from two distinct and reliable sources. In general, both data sets contain forward-looking figures for the 2 main economic indicators: annualized real GDP growth, and annualized consumer inflation rate (CPI). The figures obtained from the OECD Economic Outlook reports have semi-annual frequency and range from the first half of 1980 to the second half of 2005. The exact source of the data is the collection of paper-based OECD Economic Outlook reports, which are published two times per year (June and December). In this case, only one value is provided for each forecast, and no variability is therefore available. Whenever the forecast’s period offered is not equal to the period under analysis, a simple linear extrapolation method is used. The second source of the data is Consensus Economics Inc., one of the world's leading economic survey organizations.

10_{Austria, Belgium, Finland, France, Germany, Ireland, Italy, Luxembourg, the Netherlands, Portugal and Spain joined the EMU in}

(12)

This private firm polls each month more than 700 economists and institutions to obtain their predictions. This last feature allows the company to compute a measure of variability (standard deviation) of the forecasts at a given moment in time. In order to make this intra-temporal volatility of the forecast comparable between countries, the relative standard deviation is computed and used instead. Consensus Economics provides monthly figures that have to be altered to better represent forecasts for the desired planning periods. In order to do that, this document follows the methodology presented in Gorter, Jacobs, & de Haan (2008). For any month 𝑚 of a given year 𝑡, the 1-year forecast for a given variable will be computed as (13 − 𝑚)/12 times the forecast for year t plus (𝑚 − 1)/12 times the forecast for year 𝑡 + 1. Subsequently, these values are aggregated into quarterly figures to match the rest of the sample used throughout this document. The information obtained from this last source for the set of selected countries starts as early as the year 1989.

A special attention must be drawn to the GDPgap variable. In theory, estimates of the future output gap should be used in this alternative framework; however, neither the OECD Economic Outlook reports nor Consensus Economics publishes forecasts for the output gap variable. To handle this evident restriction, this document follows a methodology similar to Gorter, Jacobs, & de Haan (2008) to obtain a proxy of the forward-looking output gap. In this sense, a trend potential GDP growth rate is first estimated for each country in the sample, and then this value is subtracted from the expected GDP growth rate to obtain the required gap.11_{Table 2 provides}

the descriptive statistics of this trend potential real GDP growth variable.

Table 2: Descriptive Statistics. Trend Potential real GDP Growth b_{. Country List.}

Country GDP Trend Country GDP Trend

Australia 3.37 Japan 2.51

Austria 2.33 Luxembourg 4.78

Belgium 2.15 Netherlands 2.42

Canada 2.72 New Zealand 2.69

Denmark 2.06 Norway 2.95

Finland 2.83 Portugal 2.96

France 2.07 Spain 2.99

Germany 1.90 Sweden 2.51

Greece 2.14 Switzerland 1.87

Ireland 4.99 United Kingdom 2.87

Iceland 3.28 United States 3.03

Italy 1.69 - -

Average 23 countries 2.74 - -

Notes: bAnnual percentage rate.

11_{The trend potential GDP growth rate gauge is just the geometric average obtained from the historical real GDP growth series}

(13)

The final part of this paper’s analysis intends to make use of a recently presented classification of 134 countries’ monetary policy strategies (Samarina, 2014). This unique survey classifies countries on a yearly basis into five different groups: no explicit anchor, multiple targets, inflation targeting, money growth targeting, and exchange rate targeting. Some of the main sources of information for Samarina’s (2014) classification are the International Monetary Fund (IMF) Annual Reports on Exchange Arrangements and Exchange Restrictions (AREAER), and central banks’ policy reports.

Descriptive statistics of all the relevant variables included in the models are presented in the Appendix to this document.

3. Results

3.1 Forward-Looking Central Bank - OECD

(14)

However, when the monetary authority considers information regarding what inflation and

GDPgap are expected to be 6 months into the future, results change noticeably (second

column).

The interaction term CBD*ideology loses significance and (expected) inflation is no

longer a good determinant of the interest rates. On the other hand, the size of the coefficient of (expected) output gap increases while keeping its strong statistical significance: for every extra 1 percentage point in the expected value of GDPgap, short-term interest rates increase by 0.25 points, ceteris paribus. Although the interaction term CBD*ideology is still significant, the analysis of the marginal effects of ideology suggest that the conditional IDMPC hypothesis does not hold in this case. As indicated in Table 4, at a maximum level of Central Bank Dependence, ideology will have no effect on the cost of money, which is in clear opposition to the predictions of the conditional IDMPC.

Table 3: Forward-Looking Central Bank (OECD, different horizons). Regression results.

Dependent variable: short-term nominal interest rate. Dynamic bias corrected estimator.

Baseline 6-Month 12-Month 18-Month

Ideology -0.0167 -0.0152 -0.0032 -0.0054

Central Bank Dependence (CBD) 0.1793*** 0.1111 0.2264 0.2162

CBD*Ideology -0.0636** -0.1618* -0.0807 -0.0796

(Expected) Output gap 0.107*** 0.2546*** -0.0367 -0.0426

(Expected) Inflation 0.2448*** -0.027 0.2007*** 0.2051***

Lagged Dep. Var. 0.8324*** 0.7457*** 0.7368*** 0.7690***

Fixed country effects Y Y Y Y

Fixed time effects Y Y Y Y

Periodicity Quarterly Semiannual Semiannual Semiannual

Observations 1884 267 591 590

Number of N 23 7 23 23

Notes: * significant at 10%; ** significant at 5%; *** significant at 1%. Ideology and central bank dependence interacted (normalized).

(15)

Table 4: Forward-Looking Central Bank (OECD). Marginal Effects of Government Ideology on Interest Rate (Conditional on Level of Central Bank Dependence)

6 months 12 months 18 months

Minimum CBD 0.2966* (0.170) 0.1689 (0.267) 0.1645 (0.212)

Average CBD -0.0232 (0.110) 0.0021 (0.176) 0.001 (0.164)

Maximum CBD -0.2195 (0.171) -0.1192 (0.396) -0.1198 (0.355)

Notes: Standard errors in brackets; Statistical significance: * if significant at 10%; ** if significant at 5%; *** if significant at 1%.

It is important to note that the results of these first models using forward-looking information depart from the outcome of traditional Taylor-rules. Even when the calibration of the regression coefficients by Taylor is not normative, it still desirable that these are in line with some given parameters to provide the model with important features of what today is accepted as a desirable monetary rule. In particular, the “Taylor principle” is a stability condition that requires the coefficient on inflation to be larger than one to guarantee that real interest rates will actually react to inflationary pressures (Hofmann & Bogdanova, 2012). Since the models covered in this document include an interest rate smoothing term, in order to restate all coefficients by their corresponding traditional values it will be necessary to divided them by (1 − 𝛽₁). In all these first 3 variants of model (3) this Taylor principle is not present.

3.2 Forward-Looking Central Bank – Consensus Economics

Table 5 presents the results of using the forecasts obtained from Consensus Economics to compute model (3) together with some additional information. Since data from this particular source is only available starting from the first quarter of 1989, the period of analysis should also be adjusted in the baseline model to avoid introducing a new source of variability in the comparison (second column). When the monetary authority considers the expected values of

inflation and GDPgap 12 months into the future, then the results are as presented in the last

(16)

levels of Central Bank Dependence, for each unit point increase in ideology (more to the left of the political spectrum) the interest rate will also increase in 0.235 percentage points. On the other hand, at a maximum of dependence of the central bank the interest rate will be lower the more to the left the government’s ideology is. As it looks, these results are in line with the predictions of the conditional ideologically driven monetary policy cycle (IDMPC).

Table 5: Forward-Looking Central Bank (Consensus Economics). Regression results.

Dependent variable: short-term nominal interest rate. Dynamic bias corrected estimator.

Baseline Baseline - adjusted Consensus

Ideology -0.0167 -0.0493 0.0462

Central Bank Dependence (CBD) 0.1793*** 0.2374*** 0.2052***

CBD*Ideology -0.0636** -0.0978*** -0.1044***

(Expected) Output gap 0.107*** 0.1235*** 0.0267

(Expected) Inflation 0.2448*** 0.1833*** 0.1832***

Lagged Dep. Var. 0.8324*** 0.825*** 0.818***

Fixed country effects Y Y Y

Fixed time effects Y Y Y

Observations 1884 1139 936

Number of N 23 22 21

Periodicity Quarterly Quarterly Quarterly

Range 1981/2005 1989/2005 1989/2005

Table 6: Forward-Looking Central Bank (Consensus Economics). Marginal Effects of Government Ideology on Interest Rate (Conditional on Level of Central Bank Dependence)

Minimum CBD 0.235*** (0.066)

Average CBD -0.054* (0.258)

Maximum CBD -0.124** (0.056)

(17)

effect that many of the models’ variables have on the interest rate.12_{The obtained coefficients}

on the traditional Taylor-rule variables show the correct sign, are significant at the 1% level, and seem to adhere to the Taylor Principle. Clearly, the inclusion of the volatility has an impact. The variability of the inflation forecasts at a given moment increases the effect of leftist governments on the interest rate (Vol.infl*Ideol), ceteris paribus. In the same fashion, the variability of the forecasts regarding the state of the economy decreases the effect of left-wing governments on the interest rate conditional on the values of CBD (Vol.gdp*CBD*Ideology). In this context, the marginal effects of ideology on the monetary policy variable conditional on the dependence level of the central bank are presented in Table 8. The calculations of these marginal effects have been performed considering both volatility measures are at their mean values.

Table 7: Forward-Looking Central Bank. Regression results. Dynamic bias corrected

estimator.

Consensus Consensus w/Volatility

Ideology 0.0462 0.053

Central Bank Dependence (CBD) 0.2052*** 0.285***

CBD*Ideology -0.1044*** 0.017

(Expected) Output gap 0.0267 0.156***

(Expected) Inflation 0.1832*** 0.361***

Lagged Dep. Var. 0.818*** 0.929***

Vol.infl*Ideol - 0.0035* Vol.gdp*Ideol - -0.0004 Vol.infl*CBD*Ideology - 0.0011 Vol.gdp*CBD*Ideology - -0.0058*** Vol.gdp*(Exp)Output gap - -0.0032*** Vol.infl*(Exp)Inflation - -0.0067**

Fixed country effects Y Y

Fixed time effects Y Y

Observations 936 513

Number of N 21 13

Periodicity Quarterly Quarterly

Range 1989/2005 1989/2005

12_{Table 7 only presents a selection a selection of the most interesting results. The complete output of model (4) is included in the}

(18)

Finally, the results in Table 8 indicate that no conditional IDMPC seem to be present. No traditional partisan behaviour of the monetary policy can be derived from this model as left-wing governments are expected to generate higher interest rates in most of the cases.

Table 8: Marginal Effects of Government Ideology on Interest Rate Conditional on Level of Central Bank Dependence (Variabilities of Forecasts at Mean Values)

Minimum CBD 0.3858* (0.233)

Average CBD 0.2323** (0.096)

Maximum CBD 0.0572 (0.148)

3.3 Monetary Policy Strategies

(19)

Table 9: Monetary Policy Strategies. Regression Results. Dependent Variable: Short-term Nominal Interest Rate. Dynamic Bias Corrected Estimator.

Baseline IT EXRT MT

Ideology -0.0167 0.054 -0.076 -0.0728

Central Bank Dependence 0.1793*** 0.1054* 0.3392* 0.1735**

CBD*Ideology -0.0636** -0.0781* -0.0784 -0.1258**

Output gap 0.107*** 0.1428*** 0.0744** 0.0839**

Inflation 0.2448*** 0.1222* 0.4562*** 0.026

Lagged Dep. Var. 0.8324*** 0.888*** 0.750*** 0.9113***

Number of N 23 10 16 11

Number of T 104 23 38 25

Observations 1884 388 765 579

Notes: Inflation Targeting=IT; Exchange Rate Target=EXRT; Money Target=MT. Statistical significance: * if significant at 10%; ** if significant at 5%; *** if significant at 1%. Ideology and central bank dependence interacted (normalized).

Table 10: Inflation Targeting. Impact of Government Ideology on Interest Rate (Conditional on Level of Central Bank Dependence)

Minimum CBD 0.2207** (0.091)

Average CBD 0.0934** (0.044)

Maximum CBD .-0.047 (0.075)

However, the outcome of model (2) when considering only countries under a monetary regime of Exchange Rate Targeting (EXRT) is different. Assuming a high mobility of capital, this monetary regime also helps to identify the real –de jure- level of central bank dependence (low in this case). On Table 11, the focus should only be set on the marginal effect of ideology only at average values of CBD. In this case, at an average level of CBD no statistically significant marginal effect is found. In this context, partisan theories and a partial conditional IDMPC could only be present in this model if restrictions to capital mobility or other factors make the EXRT regime less efficient in limiting the influence of governments on central banks’ actions. Table 11: Exchange Rate Targeting. Impact of Government Ideology on Interest Rate (Conditional on Level of Central Bank Dependence)

Minimum CBD 0.0914 (0.252)

Average CBD -0.1017 (0.067)

Maximum CBD -0.188* (0.075)

(20)

3.4 Sensitivity Analysis

Table 12 shows the results of removing selected countries from the analysis. Based on the exclusion criteria detailed in Section 2, several countries have been excluded from the sample in independent trials and then the resulting values are contrasted with the baseline model. In particular, Iceland (ISL) was removed due to its high levels of interest rate volatility (Table 1). In general, Iceland’s exclusion did not change the predictions of the baseline model, although the specific value of inflation’s coefficient showed a marked drop in its size. In this case, the interaction term CBD*ideology was still negative and even more statistically significant than before. The joint effect of ideology on the interest rate continued to be positive at the minimum of CBD, and negative at the highest level of Central Bank Dependence (Table A2 in the Appendix).

Table 12: Regression Results. Dependent Variable: Short-term Nominal Interest Rate. Dynamic Bias Corrected Estimator.

Baseline no ISL no GRC no PRT no ESP

Ideology -0.0167 -0.0049 0.0039 -0.0149 -0.0201

Central Bank Dependence 0.1793*** 0.1702*** 0.1521*** 0.1822*** 0.1787***

CBD*Ideology -0.0636** -0.0723** -0.0483* -0.0691 -0.058*

Output gap 0.107*** 0.1087*** 0.1290*** 01096*** 0.1035***

Inflation 0.2448*** 0.1312*** 0.3225*** 0.2491*** 0.2600***

Lagged Dep. Var. 0.8324*** 0.8917*** 0.8064*** 0.8224*** 0.8314***

Number of N 23 22 22 22 22

Observations 1884 1810 1801 1809 1808

Notes: Iceland=ISL; Greece=GRC; Portugal= PRT; Spain=ESP; Italy=ITA; Statistical significance: * if significant at 10%; ** if significant at 5%; *** if significant at 1%. Ideology and central bank dependence interacted (normalized).

However, removing other countries proved to have more interesting results. Greece (GRC), Portugal (PRT), and Spain (ESP) are among the 4 countries with highest average levels of interest rate in the sample (Table 1)13_{; hence their selection for exclusion. After removing each}

of these countries from the analysis, in every case the individual effect of government ideology on the interest rate and the interaction between ideology and CBD turned not statistically significant (last 3 columns of Table 12). Inflation, output gap, and the 1-period lagged dependent variable are still able to explain the behaviour of the central bank within the context of the proposed Taylor-rule. Furthermore, the Taylor principle still holds across all the

13_{Italy (ITA) presents the third largest average level of short-term interest rate, but its exclusion from the analysis did not have}

(21)

different specifications. More interestingly, the marginal effect of ideology on the monetary policy variable loses relevance when each of these countries is excluded. Table 13 presents a clear example of this fact. After excluding Greece, government ideology is no longer able to affect the short-term interest rate, regardless of the value of CBD.

Table 13: Results Excluding Greece. Marginal Effects of Government Ideology on Interest Rate (Conditional on Level of Central Bank Dependence)

Minimum CBD 0.1071* (0.066)

Average CBD 0.0054 (0.252)

Maximum CBD -0.0655 (0.066)

These results objectively show that the presence of a conditional ideologically driven monetary policy cycle hypothesis is very sensitive to the inclusion (exclusion) of specific countries into (out of) the analysis.14

4. Final Remarks

The idea that short-term interest rates should be lower under left-wing administration only when central banks are not independent has been shown to be not robust enough across different theoretical specifications. When a forward-looking central bank is considered, then the influence of the government’s ideology on monetary policy is conditional on the source of the data. Considering forecasted figures from OECD and planning horizon equal or longer than a year resulted in no conditional ideologically driven monetary policy cycle (IDMPC) at all. If the forecasts of inflation and GDP used are the ones from Consensus Economics, then on average the effect of ideology on interest rates is the opposite as predicted by the partisan theories. At higher levels of CBD, there is still a partisan effect. The inclusion of the variability of the forecasts also causes the conditional IDMPC effect to disappear, and this model rejects the possibility of a partisan effect at all levels of central bank dependence.

Some puzzling results are also found using a backward-looking monetary authority. The paper has shown that results are inconclusive when the difference between de jure and de facto central bank independence is acknowledged. Interest rates are higher under left-wing administrations conditional on a more independent central bank only when the monetary

(22)

regime is that of Inflation Targeting. When the regime considered is that of an Exchange Rate Targeting (EXRT), the results are not aligned with the conditional IDMPC hypothesis. The latter result may contradict this hypothesis although at the same time it may be pointing out to factors reducing the autonomy of the central bank under EXRT regimes (e.g. factors limiting the mobility of capital).

Another important implication of the conditional IDMPC hypothesis is that the presence and magnitude of this effect should be the homogeneous for all countries included in the sample. However, this technical implication has also been proven to lack some robustness. In this sense, the sensibility analysis performed pointed out that this effect is highly sensitive to the inclusion or exclusion of countries into or out of the sample.

After incorporating these innovations to the analysis of the conditional ideologically driven monetary policy cycle this author is convinced that this hypothesis is only viable under certain and specific model specifications. To further strengthen this last idea, a few aspects of this paper should be put under examination. Although Taylor-rules are only one of several monetary rules’ specifications, still the violation of the Taylor principle in some cases should be analysed in more detail. Finally, the results obtained in the case of Exchange Rate Targeting regime have been at odds with the initial expectation and the source of this discrepancy should be looked into. These tasks are left as a challenge for future research.

5. Acknowledgements

(23)

6. References

Alesina, A., 1987."Macroeconomic Policy in a Two-Party System as a Repeated Game," The

Quarterly Journal of Economics, MIT Press, vol. 102(3), pages 651-78, August.

Alesina, A., Sachs, J., 1988."Political Parties and the Business Cycle in the United States, 1948-1984," Journal of Money, Credit and Banking, Blackwell Publishing, vol. 20(1), pages 63-82, February.

Alesina, A., Summers, L., 1993. "Central Bank Independence and Macroeconomic Performance: Some Comparative Evidence," Journal of Money, Credit and Banking, Blackwell Publ. Amato, J., Gerlach, S., 2002. "Inflation targeting in emerging market and transition economies:

Lessons after a decade," European Economic Review, Elsevier, vol. 46(4-5), pages 781-790, May.

Arellano, M., Bond, S., 1991. “Some tests of speci_cation for panel data: Monte Carlo evidence and an application to employment equations”. Review of Economic Studies 58, pages 277-297.

Arnone, M., Laurens, B., Sommer, M., Segalotto, J-F., 2007. "Central Bank Autonomy: Lessons from Global Trends,". IMF Working Papers 07/88, International Monetary Fund.

Belke, A., Potrafke, N., 2012. "Does government ideology matter in monetary policy? A panel data analysis for OECD countries," Journal of International Money and Finance, Elsevier, vol. 31(5), pages 1126-1139.

Blinder, A., Ehrmann, M., de Haan, J., Fratzscher, M., Jansen, D., 2008. "Central Bank communication and monetary policy: a survey of theory and evidence," Journal of

Economic Literature, American Economic Association, vol. 46(4), pages 910-45,

December.

Blundell, R., Bond, S., 1998. “Initial conditions and moment restrictions in dynamic panel data models”. Journal of Econometrics 87, pages 115-143.

Boix, C., 2000. “Partisan Governments, the international economy, and macroeconomic policies in advanced nations, 1960–93”. World Politics, vol 53, pages 38–73.

Bruno, G., 2005. "Estimation and inference in dynamic unbalanced panel-data models with a small number of individuals," Stata Journal, StataCorp LP, vol. 5(4), pages 473-500, December.

Bun, M., Kiviet, J., 2003. “On the diminishing returns of higher order terms in asymptotic expansions of bias”. Economics Letters 79, pages 145-152.

Clarida, R., Galí, J., Gertler, M., 1998. "Monetary policy rules in practice: some international evidence". European Economic Review, vol. 42, pages 1033–1067.

Clark, W., 2003. “Capitalism, Not Globalism – Capital Mobility, Central Bank Independence, and the Political Control of the Economy”, Ann Arbor: The University of Michigan Press. Cukierman, A., 2007. ”De Jure, De Facto, and Desired Independence: The Bank of Israel as a

Case Study”, in Liviatan N. and H. Barkai (eds), The Bank of Israel, Vol. II:Selected Topics in Israel’s Monetary Policy, Oxford University Press.

English, B., Nelson, W., Sack, B., 2002. "Interpreting the significance of lagged interest rate in estimated monetary policy rules". Finance and Economics Discussion Series 2002-24, Board of Governors of the Federal Reserve System (U.S.).

(24)

Gorter, J., Jacobs, J., de Haan, J., 2008."Taylor Rules for the ECB using Expectations Data,"

Scandinavian Journal of Economics 110(3), 473-488.

Grilli, V., Masciandaro, D., Tabellini, G., 1991. “Political and Monetary Institutions and Public Financial Policies in the Industrial Countries,” Economic Policy, Vol. 13, pp. 341-92 Hibbs Jr., D.A., 1977. "Political parties and macroeconomic policy". American Political Science

Review, vol. 71, pages 1467–1487.

Hofmann, B., Bogdanova, B., 2012. "Taylor rules and monetary policy: a global ‘Great Deviation’?," BIS Quarterly Review, Bank for International Settlements, September. Klomp, J., de Haan, J., 2009."Central bank independence and financial instability," Journal of

Financial Stability, Elsevier, vol. 5(4), pages 321-338, December.

Middeldorp, M., 2011. "Transparency, the Accuracy of Professional Forecasts, and Interest Rate Volatility," Federal Reserve Bank of New York, Staff Reports. Report no. 496 , May. Nickell, S. J. 1981. “Biases in dynamic models with fixed effects”. Econometrica 49, pages

1417-1426.

Orphanides, A., 2001. "Monetary Policy Rules Based on Real-Time Data." American Economic

Review, 91(4): 964-985.

Plantier, C., 2002.”The appropriate time horizon for monetary Policy”. Article published in “The policy targets agreement: Reserve Bank briefing note and related papers”. September 2002. Economics Department, Reserve Bank of New Zealand.

Potrafke, N., 2009. "Did globalization restrict Partisan politics? An empirical evaluation of social expenditures in a panel of OECD countries". Public Choice 140, 105–124.

Roger, S., Stone, M., 2005.“On Target: The International Experience with Achieving Inflation Targets,” IMF Working Paper 05/163.

Rose, A., 2007. "A stable international monetary system emerges: Inflation targeting is Bretton Woods, reversed," Journal of International Money and Finance, Elsevier, vol. 26(5), pages 663-681, September.

Roubini, N. , 2005. “Why Central Banks Should Burst Bubbles”. Working Paper, Stern School of Business, NYU.

Rudebusch, G., 1993. "The Uncertain Unit Root in Real GNP," American Economic Review, American Economic Association, vol. 83(1), pages 264-72, March.

Samarina, A., 2014. "Monetary policy strategies: abandonment, adoption, and performance," PhD Thesis. University of Groningen, Groningen, The Netherlands.

Sakamoto, T., 2008. "Economic Policy and Performance in Industrial Democracies – Party Governments, Central Banks and the Fiscal-monetary Policy Mix". Routledge, London, New York.

Taylor, J., 1993. "Discretion versus policy rules in practice," Carnegie-Rochester Conference Series on Public Policy, Elsevier, vol. 39(1), pages 195-214, December.

(25)

7. Appendix

Table A1: Baseline Model. Marginal Effects of Government Ideology on Interest Rate (Conditional on Level of Central Bank Dependence)

Minimum CBD 0.1194* (0.0631)

Average CBD -0.0167 (0.0370)

Maximum CBD -0.1077* (0.0613)

Notes: Replicated from Belke & Potrafke (2012) Standard errors in brackets; Statistical significance: * if significant at 10%; ** if significant at 5%; *** if significant at 1%.

Table A2: Results Excluding Iceland. Marginal Effects of Government Ideology on Interest Rate (Conditional on Level of Central Bank Dependence)

Minimum CBD 0.1495** (0.065)

Average CBD -0.0046 (0.025)

Maximum CBD -0.109*** (0.041)

Table A3: Wooldridge test for autocorrelation in panel data. H0: no first-order autocorrelation.

value

F(1, 22) 731.318

Prob > F 0.00000

(26)

Table A4: Complete Regression Results. Forward-Looking Central Bank With Volatility. Dynamic bias corrected estimator.

Country Consensus Consensus w/Volatility

Ideology 0.0462 0.053

Central Bank Dependence (CBD) 0.2052*** 0.285***

CBD*Ideology -0.1044*** 0.017

(Expected) Output gap 0.0267 0.156***

(Expected) Inflation 0.1832*** 0.361***

Lagged Dep. Var. 0.818*** 0.929***

Vol.infl*Lag. Dep. Var. - -0.0008

Vol.gdp*Lag. dep. var. - -0.0005***

Vol.infl*Ideology - 0.0035* Vol.gdp*Ideology - -0.0004 Vol.infl*CBD - -0.0038 Vol.gdp*CBD - -0.0062*** Vol_infl*CBD*Ideology - 0.0011 Vol_gdp*CBD*Ideology - -0.0058*** Vol.gdp*(Exp)Output Gap - -0.0032*** Vol.infl* (Exp)Inflation - -0.0067** Vol.gdp - -0.0053 Vol.infl - -0.0044

Fixed country effects Y Y

Fixed time effects Y Y

Observations 936 513

Number of N 21 13

Periodicity Quarterly Quarterly

Range 1989/2005 1989/2005

(27)

Rijksuniversiteit Groningen

Table A5: Descriptive Statistics. Variables used.

Variable Observations Mean St. Dev. Min Max

Source: OECD

Interest rate 1964 9.0 5.2 0.02 37.7

Inflation 2064 1.3 1.8 -1.7 23.3

GDPgap 2020 -0.1 1.5 -7.1 6.8

Expected inflation 6-months 308 -0.1 1.2 -5.0 3.0

Expected GDPgap 6-month 304 3.4 3.0 -2.0 18.5

Expected inflation 12-months 642 4.4 4.6 -1.9 55.0

Expected GDPgap 12-month 649 -0.4 1.3 -6.0 2.7

Expected GDPgap 18-month 644 -2.8 1.5 -10.1 -0.1

Source: Consensus E.

Expected GDPgap 12-month 961 -0.3 1.0 -3.3 2.2

Vol.infl 525 42.4 337.1 3.1 7043.6

Vol.gdp 525 29.3 48.5 3.9 494.8

Source: Others

Ideology 2064 2.9 0.9 1.00 4.0

Central Bank Dependence 1984 0.5 0.2 0.06 0.81