Does the accumulation of international reserves spur inflation? A reappraisal

North American Journal of Economics and Finance 41 (2017) 112–132

Contents lists available at ScienceDirect

North American Journal of Economics and Finance j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / e c o fi n

Does the accumulation of international reserves spur inflation? A reappraisal Andreas Steiner ⇑ ifo Institute, Poschingerstrasse 5, 81679 Munich, Germany

a r t i c l e

i n f o

Article history: Received 10 February 2016 Received in revised form 2 March 2017 Accepted 7 March 2017

JEL classification: E31 E58 F31 C23 C32

a b s t r a c t Central banks’ international reserves have increased significantly in the recent past. While this accumulation has been widely perceived as precautionary savings to prevent financial crises, rising reserves might be incompatible with the goals of monetary policy. This paper assesses the consequences for monetary policy on theoretical and empirical grounds. According to the quantity theory of money, the accumulation of reserves might result in inflationary pressures if the resulting monetary expansion is not fully sterilized and exceeds the growth of money demand. Our estimation results show that the degree of sterilization has varied considerably over time. The empirical analysis of monetary and nonmonetary determinants of inflation provides evidence that the accumulation of reserves raises the inflation rate, both on the global and the individual-country level. Ó 2017 Elsevier Inc. All rights reserved.

Keywords: International reserves Inflation Central bank policies Panel data analysis VAR

1. Introduction Central banks have significantly increased their stocks of international reserves in the recent past, especially during the first decade of the 2000s. The literature provides three main explanations for this behaviour (see Aizenman & Lee, 2007): First, the stocks of reserves might be regarded as precautionary savings intended to prevent and manage future economic crises and to mitigate output volatility (Aizenman & Ito, 2012). Second, the accumulation might be driven by a mercantilist motive: Reserve accumulation allows maintaining an undervalued exchange rate, which, in turn, promotes exports. Finally, reserve accumulation might be part of a catching-up strategy in an economy with underdeveloped financial markets where central banks assume the role of financial intermediary (Cheng, 2015). These approaches coincide in emphasizing the benefits that are associated with reserves. The costs of reserve holdings and the risks embodied in their accumulation are only marginally addressed. Whereas some studies note that reserve holdings have an opportunity cost1, the risks of reserve accumulation for price stability are generally neglected. This article aims at filling this gap. It empirically examines the inflationary consequences of reserve accumulation. Its contribution to the literature is threefold: First, our VAR analysis at the global level over the period 1970–2012 shows that global ⇑ Address: University of Groningen, Nettelbosje 2, 9747 AG Groningen, The Netherlands. 1

E-mail address: [email protected] For a discussion of various estimates of the opportunity cost of international reserves see Hauner (2006), IMF (2010a), Levy Yeyati (2008) and Rodrik (2006).

http://dx.doi.org/10.1016/j.najef.2017.03.007 1062-9408/Ó 2017 Elsevier Inc. All rights reserved.

A. Steiner / North American Journal of Economics and Finance 41 (2017) 112–132

113

reserve growth Granger-causes world inflation. Besides the focus on the more recent time period, this is a major methodological innovation compared to Heller (1976) and Khan (1979), who use simple OLS. Second, on the level of individual countries, our fixed effects panel data regressions of up to 120 countries provide a more general picture than existing studies that focus on small groups of countries. We demonstrate that reserve changes have been the major source of changes in the monetary base since the 1980s. The growth of central bank assets, in turn, is significantly correlated with inflation. Finally, we show that the degree of sterilization has varied considerably over time. In sum, these findings provide evidence for the view that reserve accumulation has inflationary consequences. To the best of our knowledge, this variety of approaches adds a more general perspective to the existing literature. Reserve accumulation might cause inflation through the following transmission channel: A rise in reserves increases the monetary base as long as it is not fully sterilized. The increased monetary base, in turn, leads through the functioning of the money multiplier to an expansion in the total amount of money. Finally, according to the quantity theory of money, the growth of money causes prices to rise after some delay. This argument is due to Heller (1976) and was empirically confirmed by Khan (1979). Since these seminal papers, a reconsideration of the relationship between reserves and inflation is overdue. Even though the risks of reserve accumulation for monetary policy have been recognized (see for example ECB, 2006)2, a theoretical and empirical study on the topic is missing. Since the studies of Heller and Khan have been published, the international financial system has changed fundamentally and their conclusions of the Bretton Woods period might not be valid in a financial system with flexible exchange rates and a high degree of capital mobility. It is therefore warranted to reconsider whether the current accumulation of reserves creates inflationary pressures. The global financial crisis has revived the debate on central banks’ objective. Their role in stabilizing the domestic and global financial system has been emphasised and their narrow focus on inflation criticised (see, among others, Eichengreen et al., 2011; Leeper & Nason, 2014; Smets, 2014). Given that central banks are equipped with one instrument only, there might be situations where the goals of low inflation and financial stability cannot be achieved jointly. This paper highlights such a dilemma of central banking: While the recent reserve accumulation is partly due to concerns for financial stability in a financially globalised world (see Obstfeld, Shambaugh, & Taylor, 2010), this policy might run counter to price stability. The uncertainty with respect to the inflationary consequences of the ongoing reserve accumulation is amplified by the fact that recent empirical studies disagree in the assessment of the extent of de facto sterilization (see the literature review in the next section). If the accumulation of reserves is not fully sterilized, it is expected to produce inflation. This might require a change in central banks’ sterilization policy since their goals of maintaining price stability and preventing currency crises via reserve accumulation can only be attained simultaneously if the effects on the monetary base are sterilized. Moreover, this policy is costly because high and variable inflation rates are found to be detrimental for growth (see Barro, 1997; Ghosh & Philips, 1998) and consumption (Assibey-Yeboah & Mohsin, 2014). Given that inflation rates might react with a significant time lag (see Batini & Nelson, 2001), there exists the risk that the negative effects of reserve accumulation are not sufficiently accounted for. This article is organized as follows: The next section summarises the existing empirical findings on the relationship between reserve accumulation, global liquidity and inflation. Section 3 reviews the theoretical links between reserves, the monetary base and inflation. Section 4 investigates this relationship empirically: First, we present a VAR analysis for world variables and, second, examine the relationship at the country level in a cross-country panel data set. The final section concludes. 2. Literature review The seminal paper exploring the links between changes in international reserves and inflation is due to Heller (1976). According to his hypothesis of global monetarism, the world price level is affected by world reserves if currencies are linked through fixed exchange rates. A regression analysis covering the period 1958–1975 (Heller, 1979) shows that world prices react with a mean lag of three years to changes in international reserves. This lag tends to be much shorter in developing than in industrial countries. A one per cent increase in reserves is estimated to result in a cumulative price increase of about 0.4 per cent. This effect is larger in developing than in industrial countries. Khan (1979) challenges these results arguing that reversed causality might be present, namely that nominal international reserves respond to inflation. However, Khan’s causality tests confirm Heller’s results for the entire period (1957–77). Conversely, tests for the floating rate period (1973–77) show that the two series are independent and that the relationship between them is rather contemporaneous than causal. Rabin and Pratt (1981) question the generality of Heller’s results. They argue that his results are driven by a short episode of reserve accumulation and rising inflation in the 1970s. According to global monetarism, inflation is a global monetary phenomenon under fixed exchange rates. Under floating exchange rates, however, the central bank can pursue an independent monetary policy such that the inflation rate is

2 The authors point out that the accumulation of reserves might challenge monetary policy: ‘‘Particularly in the presence of continuous net capital inflows, intervention on currency markets [. . .] may lead to an excessive easing of domestic monetary conditions which could then threaten price stability” (ECB, 2006, p.36).

114

A. Steiner / North American Journal of Economics and Finance 41 (2017) 112–132

determined domestically. While fixed exchange rates theoretically make foreign currency a perfect substitute for domestic currency, flexible rates eliminate this interdependence of monetary policies. There is, however, increasing empirical evidence that despite the general move to more flexible exchange rate regimes since the demise of the Bretton Woods system, global factors remain important determinants of domestic inflation rates. The persistent effect of global factors might be explained by currency substitution, endogenous money demand under exogenous central bank rates and by the violation of the uncovered interest rate parity. For a more detailed discussion of these points refer to Belke and Rees (2014) and Belke, Beckmann, and Czudaj (2014a). Finally, there is a small literature that examines the relationship between reserve accumulation and inflation at the individual country level or in a panel of countries. Findings are inconclusive: Insignificant results are reported by Ndikumana and Elhiraika (2007) who examine 21 African countries in a panel setting, by Abdullateef and Waheed (2010) who study the relationship in Nigeria using OLS and by Shrestha (2016) who analyses five South Asian countries in a panel VAR. Lin and Wang (2009) identify a positive effect of reserve accumulation on inflation in Korea and Taiwan. Using the allocation of additional Special Drawing Rights to 189 IMF countries as a natural experiment Chitßu (2016) finds the level of reserves to be inflationary. The effect, however, is indirect in the sense that a high level of reserves induces countries to pursue less prudent monetary policies because of moral hazard. For East Asian countries there is also evidence of reversed causality, namely that monetary policy reacts to the level of reserves (Shrestha & Semmler, 2015). The inflationary consequences of reserve accumulation depend crucially on the central bank’s sterilization efforts. Whereas Aizenman and Glick (2009) find that the extent of sterilization has risen in the aftermath of the Asian financial crisis, Reinhart and Reinhart (2008) show that sterilization has been decreasing during the first years of the 2000s. Besides these global approaches, another strand of the literature evaluates the sterilization efforts in individual countries. Lavigne (2008) computes sterilization coefficients of emerging markets for the 1990s and the 2000s. A comparison of both periods does not reveal a common pattern: whereas some countries increasingly sterilized reserve accumulation, others are characterised by decreasing degrees of sterilization. For a group of eight Asian countries, Ouyang, Rajan, and Willett (2008) find almost full sterilization between 1995 and 2005, with the exception of the period 2000–2002 when reserve accumulation had a positive impact on the monetary base. According to Ouyang, Rajan, and Willett (2010) China sterilized almost all reserve inflows until 2006. Since 2007, however, the degree of sterilization has fallen. Inflation may be considered a global phenomenon if the international comovement of inflation rates is high. For 22 OECD countries Ciccarelli and Mojon (2010) show that changes in their average inflation explain 70% of the country-individual inflation variance. In a follow-up study Mumtaz and Surico (2012) find that synchronised inflation rates are related to comovements in money growth rates. Neely and Rapach (2011) demonstrate that changes in global inflation explain 34% of domestic inflation variability. Spencer and Liu (2010) document empirical evidence for a common inflation trend among the OECD countries. On theoretical grounds Henriksen, Kydland, and Šustek (2013) derive synchronised inflation rates across countries from productivity growth in a business cycle model with technological spillovers. A related literature considers the inflationary consequences of global excess liquidity, defined as the amount of global liquidity that cannot be explained by a global money demand function. One possible source of global excess liquidity is the worldwide accumulation of international reserves. In particular, the IMF (IMF, 2010b) finds a strong transmission from global liquidity expansion to asset prices, which works through the accumulation of reserves. Interestingly, the transmission mechanism is quantitatively stronger for economies with fixed exchange rates. This parallels our finding that the effect of rising reserves on the price level is especially strong under fixed exchange rate regimes. Rüffer and Stracca (2006) show that global excess liquidity conveys important information about future inflation at the global level. These findings are confirmed by Belke and Rees (2014) who find global liquidity to increase inflation in a FAVAR analysis of industrialised countries. In a VAR analysis using aggregate OECD data, monetary aggregates at the global level are found to be important determinants of asset and goods prices (Belke, Orth, & Setzer, 2010). The importance of global liquidity is further documented for commodity and food prices (Belke, Beckmann, & Czudaj, 2014b; Belke, Bordon, & Volz, 2012). Berger and Harjes (2009) show that US excess liquidity Granger causes inflation in the Euro area. The mentioned studies, however, disregard the accumulation of reserves as a possible source of excess liquidity. This study shall fill this gap and examine whether inflation can partly be attributed to the growth in international reserves. 3. Reserve accumulation and inflation: theoretical aspects Since the inflationary consequences of reserve accumulation may depend on the exchange rate regime the theoretical section considers fixed and floating exchange rates separately. 3.1. Fixed exchange rates: global monetarism Standard monetarist theory states that inflation is a monetary phenomenon. Inflation can be explained by the rate of change of domestic money supply. A theoretical discussion of the relevance of money for monetary policy and inflation can be found in Woodford (2008) and Nelson (2008). Recent empirical evidence is presented by De Grauwe and Polan (2005), Thornton (2014) and Zhang (2013). Based on modern monetary aggregation theory pioneered by Barnett (1980),

A. Steiner / North American Journal of Economics and Finance 41 (2017) 112–132

115

several studies confirm the importance of money for inflation when Divisia monetary aggregates are considered (Hendrickson, 2014; Ireland, 2014). This theory is formalised by the quantity equation of money, which can be expressed in rates of change as:

^ S þ V^  Y^ p¼M where p is the inflation rate, MS the nominal money supply, V the velocity of money and Y real output. A hat on a variable denotes its rate of change. If one assumes that the velocity of money and real output growth are constant,3 money supply determines the inflation rate. Global monetarism states that under fixed exchange rates this relation also holds at the global level: The worldwide inflation rate is determined by changes in the global supply of money. Under fixed exchange rates the relative value of national currencies is stable and the world money supply, defined as the sum of domestic money supplies converted into a numeraire, is a meaningful concept. The balance of payments mechanism distributes the world money supply across countries such that the monetary market is in equilibrium in each country. If a country creates an excess supply of domestic money, its balance of payments turns into deficit: It loses reserves until the relative supply of domestic to foreign currency is again compatible with the value of the fixed exchange rate. As a result, the world money supply increases unless the foreign country sterilizes the increase in its monetary base. The link between money supply and the accumulation of international reserves can be illustrated by the money supply process. Money supply depends on two major factors: the money multiplier (m) and the monetary base (B), which equals the sum of net domestic assets (NDA) and net foreign assets (NFA):

MS ¼ m  ðNDA þ NFAÞ

ð1Þ

Hence, the rate of change of money supply can be expressed as

NDA ^S ¼ m ^ þ NFA  NFA ^ ^þ M  NDA B B

ð2Þ

Combining this specification of the money supply process with the quantity theory of money and assuming a constant velocity of money results in

^þ p¼m

NDA ^ þ NFA  NFA ^  Y^  NDA B B

ð3Þ

This equation highlights the links between changes in international reserves and the inflation rate. One can distinguish two extreme cases: If the central bank accumulates international reserves and fully sterilizes the effects on the monetary base through an open-market operation reducing its holdings of domestic bonds in exchange for money, NDA ^ ¼  NFA  N FA, ^ the price level is unaffected.4 If, however, the central bank does not sterilize ðDNDA ¼ 0Þ, the increase  N DA B

B

in international reserves directly translates into inflation. This argument holds for any individual country. If the assumptions of global monetarism are fulfilled – a world under a system of fixed exchange rates without sterilization ðDNDA ¼ 0Þ – this equation also holds for world aggregates:

^W þ pW ¼ m

NFAW BW

^ W  Y^ W NFA

ð4Þ

^ W is the growth rate of net foreign assets held by central where the superscript w denotes world variables. In particular, N FA banks aggregated across countries, that is, the growth rate of worldwide international reserves. If the world money multiplier is constant, world inflation equals the weighted rate of net foreign asset creation minus the growth rate of world output. 3.2. Floating exchange rates: national monetarism Under floating exchange rates, the central bank can pursue an independent monetary policy such that the inflation rate may vary substantially between countries. There are no monetary restrictions that make domestic inflation rates interdependent. Differences in the growth rates of the monetary base can be maintained permanently as the resulting differences in inflation rates are accommodated via changes in the exchange rate. Hence, inflation is a national phenomenon under floating exchange rates. This, however, does not foreclose that inflation rates show some co-movement due to common nonmonetary factors. With inflation rates being independent, the concept of a worldwide inflation rate determined by world money supply is no longer meaningful. Consequently, the implications of the quantity theory of money cannot be applied to the world as an aggregate. In the words of Niehans (1976, p. 179) under floating exchange rates ‘‘international reserves have lost their

3

The effect of money on the price level is a long-run effect. Over this period, output is not constant, but output growth may be close to constant. Besides open market operations, raising reserve requirements are a conventional sterilization policy. It affects money supply via a reduction of the money multiplier. 4

116

A. Steiner / North American Journal of Economics and Finance 41 (2017) 112–132

significance for inflation or deflation in the world economy, each country now controlling its own price trend.” However, for each individual country, the quantity theory still applies. Increases in the domestic (NDA) and foreign (NFA) component of the monetary base have inflationary consequences, ceteris paribus. This statement holds independently of the exchange rate regime and the degree of financial openness. The net effect of an accumulation of international reserves on the inflation rate depends on the degree of sterilization (see Eq. (3)). If countries do not offset the expansionary effect of reserve accumulation, this might raise the domestic inflation rate. If, on the contrary, they fully sterilize, the accumulation of reserves does not affect money supply and is not inflationary. To be more precise, only the direct monetary effects of the reserve accumulation are not inflationary in this case. There are, however, political economy reasons why sterilized purchases of foreign exchange might raise inflation. Since a sterilization operation expands the stock of domestic debt, policymakers might be tempted to reduce the nominal value of debt through surprise inflation. Hence, increasing domestic debt aggravates the commitment problem of the central bank and inflation expectations might rise (see Calvo, 1991). The quantity theory of money focuses on the long run, while the link between money and inflation might be weaker in the short run when prices react with a time lag. Since the accumulation of reserves might depreciate the domestic currency (see Ghosh, Ostry, & Tsangarides, 2012) before prices react, in the short run inflation might be triggered by higher import prices and turn into domestic price inflation in the longer run. 4. Empirical analysis The empirical analysis tests two hypotheses: (1) Global reserve growth drives global inflation and (2) reserve accumulation is inflationary within each individual country. The latter hypothesis is based on fewer assumptions since it does not require a fixed exchange rate regime. 4.1. Data The empirical study is carried out on the basis of a pooled data set of cross-country and time-series observations. It contains annual data from 1970 to 2012. After dropping small countries (population smaller than 3 million in the year 2000), the sample contains 123 countries, which are listed in Appendix A. Since data for several explanatory variables are missing for some countries, the number of countries used in the econometric analysis depends on the particular specification and is indicated in the respective tables. A detailed description of the variables and their data sources can be found in Appendix B. Descriptive statistics are provided in Appendix C. For the analysis of world variables the same database is used to calculate world aggregates: World reserves are defined as the sum of reserves excluding gold over all countries.5 Data on world inflation is taken from the World Development Indicators. Due to data availability it is defined on the basis of the GDP deflator instead of consumer prices. 4.2. Estimation results: global monetarism By replicating the analysis of Heller (1976) and Heller (1979), we investigate whether inflation is a global phenomenon in the sense that the growth of the stock of worldwide reserves causes global inflation. This is a time-series analysis of world data. Whereas Heller regresses the inflation rate on current and lagged changes of reserves, I apply a vector autoregression (VAR) of reserves, inflation and real income in the spirit of Estrella and Mishkin (1997) that (1) includes also lagged values of the dependent variables, (2) does not a priori assume variables to be endogenous or exogenous and (3) allows to test for the direction of causality. Thereby I can directly approach Khan’s reversed causality argument. Following the reasoning of Eq. (4), the reduced-form VAR model can be expressed as

Y t ¼ AðLÞY t1 þ CðLÞX t þ ut

ð5Þ

where Y is a k-dimensional vector of endogenous variables, X represents a nx1 vector of exogenous variables and A(L) and C (L) denote lag polynomials. u is a k-dimensional vector of reduced-form disturbances with Eðut Þ ¼ 0; Eðut u0t Þ ¼ R and Eðut u0s Þ ¼ 0 for s–t. The vector of endogenous variables includes the inflation rate (p) and the growth rate of central banks’ ^ ^ is used as an exogenous variable Xt, net foreign assets ðN FAÞ. Whereas in some specifications the growth rate of real GDP (YÞ it is included in the vector of endogenous variables in alternative specifications. As the reduced-form disturbances may be correlated contemporaneously, the reduced-form model has to be transformed into a structural model in order to identify the impact of uncorrelated structural shocks. Given the reduced-form representation and the relationship between residuals and shocks, the structural form of the VAR can be obtained by premultiplying (5) by the (kxk) matrix A0

A0 Y t ¼ A0 AðLÞY t1 þ A0 CðLÞX t þ Bet 5 Reserves are measured net of gold because reported changes in gold holdings often result from gold price changes, which affect central bank wealth but are unrelated to money supply.

A. Steiner / North American Journal of Economics and Finance 41 (2017) 112–132

117

where Bet ¼ A0 ut describes the relationship between the reduced-form disturbances ut and the structural disturbances et. By placing restrictions on A0 and B we obtain identification in the SVAR. We employ a recursive identification scheme. More precisely, we orthogonalize the reduced-form disturbances using the Cholesky decomposition. This method restricts B to a k-dimensional identity matrix and A0 to a lower triangular matrix with unit diagonal. This decomposition of the variance-covariance matrix implies a causal ordering of the variables: A shock to the first variable affects all other variables in the same time period t, whereas a shock to the second variable has a contemporaneous effect on the third variable, but passes-through to the first variable with a lag of one period. To determine the number of included lags, we run a variety of tests: Minimization of the final prediction error (FPE) and the Akaike information criterion (AIC) suggest an optimal lag length of three periods. This is confirmed by an LR-test, which considers the null hypothesis that all coefficients on the lags of the endogenous variables are zero: Starting with a maximum of 10 lags and reducing this number step by step, the hypothesis can be rejected for the first time at a level of significance of 10% when three lags are included. Results are confirmed when the Lütkepohl statistics are used, who recommends to drop the constant term from the log likelihood. When estimating the variance-covariance-matrix, we apply a small-sample degrees-of-freedom adjustment. Interpretation of a VAR requires that it is stable, that is, it is invertible and may be represented by an infinite-order vector moving-average. Our diagnosis tests show that the modulus of each eigenvalue of the companion matrix is strictly less than one, which is a sufficient condition for the VAR to be stable. As an additional test we check the null hypothesis that the disturbances are normally distributed. To this end, we compute three test statistics, namely a skewness statistic, a kurtosis statistic and the Jarque-Bera statistic. In most cases, the null hypothesis cannot be rejected. There is some indication that specifications excluding the growth rate of real GDP may be preferred. Fig. 1 shows the orthogonalised impulse-response function for an expansionary reserve growth shock on inflation based on the Cholesky decomposition. The first two panels present a bivariate VAR for inflation and reserve growth. The following two panels add the growth rate of real GDP as an exogenous determinant. All remaining graphs are based on VARs considering inflation, reserve growth and real GDP growth as endogenous variables. Individual panels are based on different orderings of the variables, which are indicated in the subtitles. Given the lack of an unambiguous theoretical ordering of the variables and as a robustness check, we report all possible orderings of the variables. Independently of the particular ordering, reserve growth increases inflation slightly in the year following the shock and unfolds its largest and significant effect in the third year. This effect only decreases slowly over time. With respect to the magnitude of the effect, a shock equal to an increase of the global stock of reserves by 10% – a one standard deviation in the growth rate of reserves- raises world inflation by 1.4 percentage points. Fig 2 shows the orthogonalised response of the growth rate of reserves to an inflationary impulse for the same set of specifications as Fig 1. There is no significant effect of inflation on the growth rate of reserves. To check the causality of this relationship, a Granger test is applied. A variable is said to Granger-cause x when its past values help to predict x. The hypothesis that the growth of reserves does not cause inflation can be rejected at the 1%-significance level across all specifications. With respect to reversed causality, we find that inflation does not Grangercause reserves to grow. The inclusion of the growth rate of real GDP does not affect our main results. There is some evidence that real GDP growth contributes to the accumulation of reserves. As discussed in the theoretical section, the conclusion of global monetarism – global reserve growth causes world inflation – is based on the assumption that exchange rates are fixed. Therefore, I replicate the VAR analysis for a subsample that contains only those annual observations for which a de facto fixed or managed exchange rate was in place.6 Based on these observations, world inflation is calculated as the mean of country-specific growth rates of the GDP deflator, which are weighted by a country’s fraction of nominal world GDP. Weights are calculated for each year individually such that changes in the relative size of an economy are accounted for. The results are presented in Fig. 3. Again, reserve growth raises inflation beginning in the first year and the effect becomes significant and strongest in the third year. The magnitude of the effect is larger than in the full sample: If reserves grow by 10%, three years later inflation will have increased by 3 percentage points. Granger tests support the hypothesis that the growth of reserves causes inflation. The null hypothesis of no effect can again be rejected at a significance level of 1%. The hypothesis of reversed causality can be rejected. Besides reserve growth, the growth of real GDP is found to Granger-cause world inflation. As a further robustness test we replicate our VAR analysis for different lag lengths. In particular, we include two and four lags instead of the three lags, which are indicated as the optimal lag length. In addition, we consider a shorter time period starting in 1995. By and large, results are confirmed. As a result, global monetarism can be confirmed: The growth in international reserves pushes global inflation upwards and this effect is stronger in countries characterised by fixed exchange rate regimes.

6 This classification is flawed for our purposes and may be sharpened in future research: Global monetarism makes its statements for countries whose currencies are fixed with respect to one common currency. The empirical group of pegged exchange rates, however, consists of various blocs of fixed exchange rates with respect to different currencies, i.e. the Dollar, Euro or Yen.

118

A. Steiner / North American Journal of Economics and Finance 41 (2017) 112–132

2

1

0

-1

2

1

0

%

-1

2

1

0

-1 0

1

2

3

4

5

0

1

2

3

4

5

2

1

0

-1 0

1

2

3

4

5

Years after shock 95% confidence interval

impulse-response (orthogonalised)

Fig. 1. Impulse-response functions: response of inflation to an expansionary reserve growth shock. Note: The graphs display the time path of the response of the inflation rate to reserve growth. The order of the variables is indicated in the subtitles. The first four panels are based on bivariate VARs including the inflation rate and the growth rate of reserves, while the remaining specifications additionally include the growth rate of real GDP as an endogenous variable. Exog. Y means that the growth rate of real GDP is included as an exogenous regressor.

4.3. Estimation results: national monetarism This section examines the hypothesis that inflation is rather a national than a global phenomenon. To this end, we first show graphically the correlation between reserve growth and the growth rate of base money (Fig 4) and reserve growth and the inflation rate (Fig 5). Each data point represents the average growth rates of both variables over the period 1970–2012 for an individual country. Whereas the left-hand side graph is based on the full sample, the graph on the right hand displays the relationship for a reduced sample excluding outliers, namely countries with a growth rate of reserves and/or a growth rate of

119

A. Steiner / North American Journal of Economics and Finance 41 (2017) 112–132

4 2 0 -2 -4

4 2 0 -2

%

-4

4 2 0 -2 -4 0

1

2

3

4

5

0

1

2

3

4

5

4 2 0 -2 -4 0

1

2

3

4

5

Years after shock 95% confidence interval

impulse-response (orthogonalised)

Fig. 2. Impulse-response functions: response of the growth rate of reserves to an inflationary shock. Note: see notes to Fig. 1.

base money or inflation, respectively, exceeding 50%. Fig 4 shows that countries’ growth rate of the monetary base is larger, the higher their growth rate of international reserves is. Fig 5 depicts a positive relationship between reserve growth and inflation. Next, we analyse individual country data in a panel framework in the spirit of Eq. (3). Because of sterilization policies, our ^ ^ are negatively correlated. Due to this multicollinearity, estimation of Eq. (3) might main explanatory variables N DAand N FA lead to imprecise coefficient estimates and large variances. Moreover, conclusions would be misleading because the coeffi^ indicated its inflationary impact for given NDA, that is, for the case of no sterilization. We therefore prefer to cient of N FA proceed by a two-step procedure: First, we examine the relationship between inflation and the entire monetary base, defined as the sum of NDA and NFA. Second, we decompose the growth of the monetary base in the contributions of NDA and NFA.

120

A. Steiner / North American Journal of Economics and Finance 41 (2017) 112–132

4 2 0 -2 -4

4 2 0 -2

%

-4

4 2 0 -2 -4 0

1

2

3

4

5

0

1

2

3

4

5

4 2 0 -2 -4 0

1

2

3

4

5

Years after shock 95% confidence interval

impulse-response (orthogonalised)

Fig. 3. Impulse-response functions: response of inflation to an expansionary reserve growth shock (subsample of fixed exchange rates). Note: see notes to Fig. 1.

In the first step, the pooled fixed effects estimator is used to estimate the following specification7:

pi;t ¼

p X

aj  B^ i;tj þ c  X i;t þ ci þ ei;t

ð6Þ

j¼0

where B is the monetary base, X a set of control variables and c a fixed country effect. Inflation p is measured by the change in the consumer price index. i and t are the country and year indices, respectively. As explained in Section 3 we do not need to 7

According to a Hausman specification test the fixed effects model is to be preferred to a random effects specification.

121

A. Steiner / North American Journal of Economics and Finance 41 (2017) 112–132

600

50

40 400 30

200

20

10 0 0

20

40

60

80

100

10

Growth rate of reserves (%) Average growth rate

20

30

40

50

Growth rate of reserves (%)

Fitted values

Average growth rate

Fitted values

Fig. 4. Reserves and monetary base. Note: Whereas the left-hand side graph is based on the full sample, the graph on the right hand displays the relationship for a reduced sample excluding outliers, namely countries with a growth rate of reserves and/or a growth rate of base money larger than 50%.

1000 40

800 30

600 20

400 10

200

0

0

0

20

40

60

80

Growth rate of reserves (%) Average growth rate

Fitted values

100

10

20

30

40

50

Growth rate of reserves (%) Average growth rate

Fitted values

Fig. 5. Reserves and inflation. Note: Whereas the left-hand side graph is based on the full sample, the graph on the right hand displays the relationship for a reduced sample excluding outliers, namely countries with a growth rate of reserves and/or an inflation rate exceeding 50%.

control for exchange rate regimes and financial openness. The inflationary effects of changes in the monetary base are independent of the exchange rate regime and capital mobility. Column 1 of Table 1 examines the bivariate relationship between the growth rate of the monetary base and the contemporaneous inflation rate. The positive and statistically significant coefficient indicates that inflation rises by 6.9 percentage points when the monetary base grows by 10%. According to the money market equilibrium condition, monetary expansion raises inflation only to the extent to which it exceeds the growth of money demand, e.g., real GDP growth. In line with Eq. (3), column 2 adds the growth rate of real GDP as an additional determinant. Its effect on inflation is negative but insignificant while the coefficient on the growth rate of the monetary base is virtually unchanged. During episodes of exceptionally high inflation rates the relationship between money and inflation might be unstable: Inflation is driven by inflation expectations, the velocity of money may increase and money creation itself may become endogenous to the inflation rate (see Sargent & Wallace, 1973). Column 3 controls for episodes of high inflation by the inclusion of two dummy variables. Following Dornbusch and Fischer (1993), inflation is considered to be high if its annual rate exceeds 50 per cent. Hyperinflations are defined as periods of annual inflation exceeding 1000 per cent. While the dummy variables are both positive and significant, the coefficient on the growth rate of the monetary base is still positive and significant, albeit with lower magnitude. The inclusion of fixed time effects (column 4), does not affect the results. Economic theory suggests that inflation reacts with some time lag to increases in money supply. Existing evidence shows that monetary factors feed into inflation with a lag of up to three years. Therefore, in columns 5–7 we additionally include growth rates of the monetary base lagged by one and two years. While the contemporaneous effect of money on inflation is larger after the inclusion of lags, lagged rates of monetary growth do not significantly affect inflation. As an alternative way to control for the dynamic relationship between money and inflation, columns 8 and 9 include the average growth rate of money during the preceding three years. Its effect is positive and significant; after controlling for high- and hyperinflations, an average growth rate of the monetary base during the preceding three years of 10% raises the inflation rate by 4.9 percentage points. These estimates are robust to the inclusion of fixed time effects in columns 10 and 11. The relationship between changes in the monetary base and inflation might be non-linear. As a robustness check we split our sample into two subsamples, one where the monetary base increases and another where the monetary base decreases. In

122

(1) Growth in B

0.6867 (2.23)

Growth in B, lagged by 1 year Growth in B lagged by 2 years Growth in B, mean over preceding 3 years Growth of real GDP Hyperinflation, dummy High inflation, dummy Observations Number of countries R-squared Time effects

(2) **

(3) **

0.6854 (2.22)

0.0389 (0.02)

3489 120 0.39 no

3323 119 0.39 no

(4) *

0.4671 (1.84)

2.9114 (0.96) 28.2577**** (3.71) 0.2774 (0.83) 3323 119 0.52 no

(5) *

0.4676 (1.84)

3.2711 (0.99) 28.0358*** (3.74) 0.2472 (0.72) 3323 119 0.52 yes

(6) ***

1.0332 (2.73) 0.0201 (1.04) 0.0622 (1.08)

1.7836 (1.18)

3141 119 0.59 no

(7) **

0.8621 (2.49) 0.0041 (0.21) 0.0608 (1.04)

0.2861 (0.24) 15.8263*** (2.93) 0.0363 (0.09) 3141 119 0.62 no

(8)

(9)

(10)

(11)

0.8226** (2.29) 4.1575* (1.92)

0.4931** (2.06) 0.3240 (0.16) 36.4637*** (3.78) 0.2485 (0.46) 3141 119 0.43 no

0.8225** (2.29) 4.4290* (1.91)

0.4937** (2.05) 0.3326 (0.15) 36.4476*** (3.77) 0.2452 (0.44) 2487 97 0.43 yes

**

0.8639 (2.50) 0.0034 (0.17) 0.0608 (1.04)

0.5481 (0.52) 15.7017*** (2.93) 0.0414 (0.10) 3141 119 0.62 yes

3141 119 0.21 no

Notes: Estimation by OLS including fixed country effects. t-statistics (in brackets) computed with heteroskedasticity-consistent standard errors. *, respectively.

**

, and

***

2487 97 0.21 yes

denote significance at the 10%, 5%, and 1% levels,

A. Steiner / North American Journal of Economics and Finance 41 (2017) 112–132

Table 1 Monetary base and inflation. Dependent variable: Inflation (CPI-based).

A. Steiner / North American Journal of Economics and Finance 41 (2017) 112–132

123

case of a growing monetary base our result of a positive relationship between money and inflation is confirmed. When the monetary base decreases, however, the relationship seems to be absent.8 Although panel unit root tests indicate that individual time series are stationary, panel cointegration tests provide evidence for a cointegration relationship between inflation, monetary growth and the growth rate of nominal GDP. Accordingly, we proceed by a cointegration analysis based on an error correction model that allows differentiating between the long-run coefficients and the adjustment process in the short run. Estimation is based on the mean group estimator as proposed by Pesaran and Smith (1995). Results, which might be obtained upon request, confirm our results from the panel data regressions. Growth in the monetary base raises inflation. In sum, in accordance with monetary theory of inflation, we find a significant positive relationship between money growth and inflation across different specifications and country samples. This result is in line with the findings of, among others, Lucas (1980), De Grauwe and Polan (2005) and Benati (2009). We therefore proceed by the second step of our analysis and examine the extent to which changes in the monetary base are due to changes in NFA. Fig. 6 depicts the time path of NFA and NDA at four selected central banks over the period 1970–2012. The difference between the US and the three emerging markets Korea, Malaysia and Mexico is striking: Foreign assets of the US Federal Reserve are relatively constant; changes in the monetary base are driven by the increase in domestic assets. In the selected emerging markets, the opposite holds: Besides some swings, domestic central bank assets are relatively constant and changes in the monetary base are due to increases in foreign assets. This is especially true since the East Asian financial crisis in 1997. Fig. 7 shows the evolution of the share of foreign assets in the total monetary base averaged across countries. Two facts are noteworthy: The share of foreign assets in total central bank assets has been increasing over time. Second, this holds for industrialised and developing countries alike. While in the 1980s and 1990s foreign assets account for a larger component in industrialised countries compared to developing ones, the shares of both country groups have converged in the 2000s. Table 2 provides information about the sources of changes in the monetary base. For each country the ratio of the change in NFA to the change in the monetary base is calculated and then averaged across countries. For changes in NDA the same procedure is applied. Changes are computed over decades. The following conclusions can be drawn: Since the 1980s changes in NFA have been the most important source of changes in the monetary base. The 1980s were an exceptional period where the increase in NFA was larger than the increase in the monetary base because sterilization policies led to a decrease in NDA. Developments in developing countries are not significantly different from the behaviour in the full sample including industrialised countries. While the preceding analysis of inflation has focused on monetary factors, Appendix D considers a broader set of inflation determinants including macroeconomic and political factors. These complementary results support the finding that reserve accumulation has inflationary effects. We conclude that the increase in international reserves has been an important determinant of inflation across countries: First, increases in international reserves are the leading factor of the growth of the monetary base. Second, increases in the monetary base are associated with higher inflation rates. 4.4. Sterilization policies As pointed out in the theoretical section, the effect of an increase in international reserves on the inflation rate depends on the degree of sterilization. If the effect of reserve accumulation on the monetary base is fully sterilized through the sale of domestic bonds, its inflationary impact is cushioned.9 To measure the extent of sterilization, I estimate sterilization coefficients. Along the lines of Aizenman and Glick (2009) I regress the change in domestic central bank assets on the change in reserves (both scaled by the monetary base) and the growth rate of nominal GDP, that is

    DNDA DNFA ^ i;t þ ei;t ¼a þ b  GDP B B i;t i;t

ð7Þ

The sterilization coefficient is defined as the coefficient a in this regression. A sterilization coefficient of minus one represents full sterilization while a coefficient of zero implies no sterilization. Nominal GDP growth is included to control for changes in nominal money demand. Hence, with full sterilization the central bank prevents an expansion of the monetary base due to changes in reserves, but domestic credit may grow due to other reasons, namely increases in money demand. The results of a cross-section analysis are presented in the upper part (Panel A) of Table 3. The estimation assumes by definition that the sterilization coefficient is constant over time and across all countries. It equals –0.57 (column 1). This indicates that the effect of reserve growth on the monetary base was partially offset by the sale of domestic central bank assets. However, the magnitude of the coefficient shows that sterilization is far from complete.

8

Detailed regression results may be obtained from the author upon request. Imperfect substitutability between assets is a precondition for the effectiveness of sterilization. Any sterilized intervention raises the domestic interest rate, which in turn attracts additional capital inflows. If assets are perfect substitutes, these capital inflows persist until the interest rate differential has disappeared. 9

124

A. Steiner / North American Journal of Economics and Finance 41 (2017) 112–132

USA

Korea, Rep.

1500

300

1000

200

500

100

0

0 1970

1980

1990

2000

2010

1970

1980

Year Domestic assets

1990

2000

2010

Year Foreign assets

Domestic assets

Malaysia

Foreign assets

Mexico

150 150

100

100

50

50

0

0 1970

1980

1990

2000

2010

1970

1980

Year Domestic assets

1990

2000

2010

Year Foreign assets

Domestic assets

Foreign assets

Fig. 6. Components of central bank balance sheets. Note: This figure shows the evolution of the balance sheet components of four selected central banks. Central bank assets consist of domestic assets or refinancing credit and foreign assets. The latter equal international reserves.

1.00

0.75

0.50

0.25

0.00 1970

1980

1990

2000

2010

Year Full sample

Industrialised countries

Fig. 7. Share of foreign assets in monetary base. Note: This figure shows the share of net foreign assets in the total monetary base averaged across countries. The monetary base is defined as the sum of net domestic assets and net foreign assets.

Sterilization policies of central banks might change over time. They depend on the primary objectives of central bank policy, economic conditions and the policies of other central banks. To investigate the stability of the relationship over time, I estimate the regression for different time periods. The sample is divided into three time periods where breaks are defined by important financial crises: the debt crisis in 1982 and the East Asian financial crisis in 1997. Columns 2–4 of Table 3 present the results. The sterilization coefficient amounts to 0.88 in the 1970s, increases to 0.36 between 1983 and 1996 and falls again to 0.48 between 1997 and 2012. This indicates that sterilization has been present in all periods, albeit to different extents. The time-varying feature of reserve accumulation and its sterilization might be captured by a fixed effects panel data analysis, which uses each individual data point instead of averaged data. The results, which are presented in panel B of Table 3, confirm that central banks partially sterilize changes in the foreign component of their monetary base. The degree of sterilization, however, is lower compared to the cross-section analysis, namely 33%. It has been decreasing over time.

A. Steiner / North American Journal of Economics and Finance 41 (2017) 112–132

125

Table 2 Source of changes in the monetary base. Full sample

1970–79 1980–89 1990–99 2000–10

Developing countries

NFA (1)

NDA (2)

NFA (3)

NDA (4)

0.35 1.87 0.65 0.80

0.65 0.87 0.35 0.20

0.35 2.28 0.60 0.79

0.65 1.28 0.40 0.21

Note: Numbers are calculated as the ratio of the change in NFA or NDA over the change in the monetary base.

Table 3 Sterilization. Dependent variable: Change in domestic central bank assets (relative to monetary base). Panel A: Estimation method: Cross-section analysis (between regression) (1) 1970–2012 Change in international reserves (relative to monetary base) Nominal GDP (growth rate) Observations R-squared Number of countries Panel B: Estimation method: Panel data analysis

Change in international reserves (relative to monetary base) Nominal GDP (growth rate) Observations R-squared Number of countries

(2) 1970–1982

0.8773*** 0.5715*** (6.08) (9.61) 0.0332*** 0.2707*** (8.63) (14.03) 3794 999 0.46 0.77 123 89 including fixed country effects (1) (2) 1970–2012 1970–1982 0.3327*** (10.86) 0.0056** (2.34) 3794 0.13 123

0.4060*** (6.97) 0.1082*** (8.23) 999 0.25 89

(3) 1983–1996

(4) 1997–2012

0.3600*** (3.01) 0.0112*** (6.62) 1341 0.28 118

0.4777*** (4.69) 0.1968*** (5.36) 1454 0.25 121

(3) 1983–1996

(4) 1997–2012

0.3743*** (7.05) 0.0037* (1.78) 1341 0.12 118

0.2289*** (7.80) 0.0529*** (4.22) 1454 0.09 121

Notes: t-statistics (in brackets) computed with heteroskedasticity-consistent standard errors. *, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively.

Interest rates affect the demand for money and might be an additional determinant in the sterilization regressions. We therefore replicate the analyses of Table 3 and add the change in the interest rate – measured by the deposit interest rate – as control variable. Our results are robust to this variation: Sterilization coefficients are negative and significant in all time periods and have decreased over time.10 Given this instability of sterilization coefficients, it might be appropriate to focus on shorter time periods. To this end, I use a rolling regression over a period of five years. The first regression covers the period from 1970 to 1974, the second from 1971 to 1975 and the last uses data between 2008 and 2012. Fig 8 plots the sterilization coefficients. Sterilization increases after the end of the Bretton Woods system. The effect of reserve changes on the monetary base are at least cut by half. After low degrees of sterilization in the aftermath of the debt crisis of 1982, sterilization further increases until the late 1980s but then falls abruptly to relatively low levels. During the 2000s sterilization remains low, both in absolute terms and in comparison with other periods. An explanation for the decreasing sterilization efforts of central banks might be given by the increase of capital mobility. Private capital flows may render central banks’ sterilization policies ineffective. If a central bank sterilizes its accumulation of reserves, the interest rate is higher than without sterilization. Hence, foreign capital flows might continue and challenge the central bank to further accumulate reserves, thereby offsetting its sterilization policies. Finally, the degree of sterilization might depend on the exchange rate system. It is expected that sterilization is lower in flexible exchange rate regimes compared to fixed ones. Nevertheless, if a central bank pursues the goal to accumulate reserves without affecting the exchange rate, it has to sterilize its foreign exchange accumulation even under floating exchange rates. To test whether sterilization coefficients differ systematically between fixed and floating exchange rate systems, we split the dataset in two subsamples. Results, which may be obtained from the author upon request, do not point to significant differences between both subsamples.

10

Results may be obtained from the author upon request.

126

A. Steiner / North American Journal of Economics and Finance 41 (2017) 112–132

0.00

-0.20

-0.40

-0.60

-0.80

-1.00

-1.20 1970

1975

1980

1985

1990

1995

2000

2005

2010

Year

Fig. 8. Sterilization coefficients. Note: This graph is based on rolling regressions of the rate of growth of domestic central bank assets on the growth rate of international reserves. A total of 39 regressions is estimated, each covering 5 years. The graph plots the resulting coefficients attributed to international reserve growth. The year on the x-coordinate marks the last year of the respective regression, e.g. the coefficient in 1975 is based on the regression over 1971–1975.

5. Conclusions This article examines the inflationary consequences of reserve accumulation, both at the global and the country level. First, the global level of reserves affects world inflation. Second, this relationship is confirmed to hold at the country level alike. At the global level, a rate of reserve growth of 10% causes an increase of the price level of 1.4 percentage points over a period of three years ceteris paribus. This is an economically significant effect. Reserve growth increases inflation slightly in the year following the shock and unfolds its largest and significant effect in the third year. This effect only decreases slowly over time. This result is robust to the choice of lag length, variable ordering in the VAR and specific time period. With respect to causality, global reserve growth is found to Granger-cause inflation. There is no evidence that reserve policies react to inflation. In a subsample of countries operating under fixed exchange rate systems, the effect is twice as large: A rate of reserve growth of 10% causes an increase of the price level of 3 percentage points over a period of three years. On the level of individual countries, reserve changes have been the major source of changes in the monetary base since the 1980s. Our panel data analysis shows that the growth of central bank assets, in turn, is significantly correlated with inflation. If the monetary base increases by 1%, inflation rises during a period of three years on average by 0.72 percentage points.11 Taken together, reserve increases are accountable for rising price levels via their expansionary effect on money supply. This finding is robust: It persists after adding real GDP growth and dummy variables for high inflation episodes as controls. It is independent of the chosen dynamic structure of how changes in the monetary base feed into inflation. To account for potential cointegration relationships we also conduct a cointegration analysis based on an error correction model, which confirms significant positive effects of the growth in the monetary base on inflation. Finally, this finding is confirmed in a panel data analysis of the non-monetary determinants of inflation. After controlling for a large number of economic, political and institutional variables, we find that a policy of reserve accumulation positively affects inflation. To complete the picture of the relationship between monetary growth, reserve accumulation and inflation we estimate sterilization coefficients. To check for the robustness of our results we use different estimation methods, add controls for GDP growth and interest rates and consider shorter time periods. Results show that reserve accumulation is sterilized only partly and that the degree of sterilization varies considerably over time. By tendency sterilization has become weaker over time: While it was relatively strong in the 1970s, its low degree during the 2000s might be due to reduced effectiveness in a financially integrated world. On the other hand, since the recent accumulation of reserves is sterilized to a lower extent than in previous periods, its inflationary impact might be larger in the long run. In sum, the result of Heller (1976) that the accumulation of reserves has inflationary effects can be confirmed both at the global and the country-specific level. This shows that central banks face a trade-off: Their policy of reserve accumulation, which is intended to increase financial stability in the face of external shocks, endangers another central bank objective, namely price stability.

11

The effect is calculated after averaging the coefficient on the monetary base across all specifications in Table 1.

A. Steiner / North American Journal of Economics and Finance 41 (2017) 112–132

127

Acknowledgements The author thanks two anonymous referees, Michel Normandin, Carsten Trenkler, Roland Vaubel, Frank Westermann, Joachim Wilde and participants at the World Congress of the Econometric Society, the International Panel Data Conference and the ASSA meeting for helpful comments and suggestions. Appendix A Country list. Albania Algeria Angola Argentina Armenia Australia Austria Azerbaijan Bangladesh Belarus Belgium Benin Bolivia Bostswana Brazil Bulgaria Burkina Faso Burundi Cambodia Cameroon Canada Central African Republic Chad China Colombia Congo, Dem. Rep. Congo, Rep. Costa Rica Cote d’Ivoire Croatia

Czech Republic Denmark Dominican Republic Ecuador Egypt, Arab Rep. El Salvador Ethiopia Finland France Georgia Germany Ghana Greece Guatemala Guinea Haiti Honduras Hungary India Indonesia Iran, Islamic Rep. Iraq Ireland Israel Italy Japan Jordan Kazakhstan Kenya Korea, Rep. Kyrgyz Republic

Lao PDR Lebanon Libya Lithuania Madagascar Malawi Malaysia Mali Mexico Moldova Morocco Mozambique Nepal Netherlands New Zealand Nicaragua Niger Nigeria Norway Pakistan Panama Papua New Guinea Paraguay Peru Philippines Poland Portugal Romania Russian Federation Rwanda Senegal

Serbia Sierra Leone Singapore Slovak Republic South Africa Spain Sri Lanka Sudan Sweden Switzerland Syrian Arab Republic Tajikistan Tanzania Thailand Togo Tunisia Turkey Uganda Ukraine United Arab Emirates United Kingdom United States Uruguay Venezuela, RB Vietnam Yemen, Rep. Zambia Zimbabwe

Appendix B List of variables and data sources. Variable

Source

Description

Net foreign assets (NFA)

WDI

NFA are measured by net international reserves, which comprise special drawing rights, reserves of IMF members held with the IMF, and holdings of foreign exchange under the control of monetary authorities. Gold holdings are excluded Domestic central banks assets are defined as the sum of the following seven items: claims on central government, claims on deposit money banks, claims on state and local governments, claims on nonfinancial public enterprises, claims on the private sector, claims on other banking institutions and claims on nonbank financial institutions. The

Domestic central IFS bank assets (NDA)

(continued on next page)

128

A. Steiner / North American Journal of Economics and Finance 41 (2017) 112–132

Appendix B (continued)

Variable

Source

Description

variable is set to missing if all seven items are missing Sum of NFA and NDA WDI In panel data analysis: Inflation is measured by the annual percentage change in the consumer price index In world data analysis: Inflation is measured as the growth rate (in%) of the GDP implicit deflator which is defined as the ratio of GDP in current local currency to GDP in constant local currency Real GDP WDI GDP in constant local currency units De jure fixed Ghosh, Gulde, and Wolf (2002) Equals one if one of the following finer categories applies: exchange rates, and own update based on AREA- dollarized, currency board, monetary union, single currency dummy ER peg, published basket peg and secret basket peg De facto fixed and Ilzetzki, Reinhart, and Rogoff Dummy variable equals one if exchange rate code equals 1, 2 managed exchange (2010) or 3 in the coarse classification rate Trade openness WDI Sum of exports and imports divided by GDP Capital account Chinn and Ito (2006) and update Measure of the de jure openness of the capital account openness (kaopen). Calculation is based on the binary dummy variables of the IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER) Public debt Abbas, Belhocine, El-Ganainy, Gross general government debt or, where not available, and Horton (2010) central government debt relative to GDP Domestic credit WDI Domestic credit to private sector (% of GDP) Democracy Marshall and Jaggers (2013) Democracy is measured by a score, which combines the information contained in indicators of democracy and autocracy (POLITY2 variable). It ranges from +10 (strongly democratic) to-10 (strongly autocratic) Financial crisis Reinhart and Rogoff (2011) Dummy variable that takes on the value of one if at least one of the following crises is identified: banking crisis, currency crash or sovereign debt crisis (external or domestic)

Monetary base (B) Inflation

Data sources: IFS: International Financial Statistics, IMF; WDI: World Development Indicators, World Bank.

Appendix C Descriptive statistics. Variable

Full sample Obs. Mean

Inflation (GDP based) Inflation (CPI) Highinflation (CPI) Hyperinflation (CPI) Growth in B Growth in B, mean over 3 years Growth of real GDP Change NDA/B Change in int. reserves/B Growth of nominal GDP Fixed exchange rate, dummy Trade openness

Sample of moderate inflation rates Std. dev. Min.

Max.

Inflation measures 4752 54.08 584.77 32.81 26762 4226 47.16 599.90 17.63 24411 4226 0.0637 0.2442 0 1 4226 0.0069 0.0826 0 1 Monetary base and inflation (Table 1) 3948 0.5856 6.3982 0.9995 219.25 3690 0.5398 4.1254 0.3831 145.21 4714 0.0356 0.0577 0.5581 0.8896 Sterilization (Table 3) 3954 0.0833 0.1719 0.9953 1.00 3963 0.0524 0.1486 0.9112 0.9827 4802 0.5491 5.3524 0.4922 257.14 Determinants of inflation (Table D) 4766 0.4778 0.4996 0 1 4727 69.23 47.76 0.3088 447.24

Obs. Mean

Std. dev. Min.

Max.

3719 9.4564 20.25 3957 9.0887 9.2033

32.81 17.63

3258 0.1862 0.3819 3079 0.2613 1.9405 3699 0.0399 0.0454

0.9995 13.95 0.3831 69.28 0.4130 0.4650

3258 0.0582 0.1361 3264 0.0541 0.1433 3743 0.1388 0.1769

0.9452 0.9296 0.9112 0.8487 0.3403 8.0196

3832 0.4611 0.4986 3712 70.59 50.13

0 0.3088

1065 50

1 447.24

129

A. Steiner / North American Journal of Economics and Finance 41 (2017) 112–132 Appendix C (continued)

Variable

Full sample Obs. Mean

Capital account openness World Inflation Public debt GDP per capita Domestic credit Democracy F. reserves, growth (weighted) F. reserves, growth (mean o. 3 yrs.) Financial crisis

4576 5719 4316 4805 4451 5073 3970 3711 2743

0.0371 15.49 58.79 7868 41.24 1.1076 0.1667 0.1594 0.3966

Sample of moderate inflation rates Std. dev. Min. 1.5333 9.6619 54.48 12213 40.98 7.3195 1.1743 0.6365 0.4893

1.8640 4.2080 0 113.01 0.5574 10 0.4383 0.2285 0

Max.

Obs. Mean

Std. dev. Min.

2.4390 48.58 2093 81947 233.55 10 58.39 20.50 1

3690 3957 3589 3739 3710 3889 3268 3091 2207

1.5307 9.3948 36.65 12599 43.04 7.1213 0.7512 0.6832 0.4761

0.0843 14.53 55.78 8647 44.24 2.3096 0.1479 0.1649 0.3471

1.8640 4.2080 0 113.01 0.6877 10 0.4383 0.2285 0

Max. 2.4390 48.58 257.93 67805 233.55 10 28.09 20.50 1

Appendix D Non-monetary analysis of inflation determinants The preceding sections were based on the conjecture that inflation is a monetary phenomenon. Accordingly, we explained changes in money supply directly by monetary factors, that is, on the basis of the central bank balance sheet. This, however, raises another question: What are the underlying reasons for inflationary increases in money supply? To answer this question, this appendix presents an alternative analysis focusing on non-monetary factors that might explain central banks’ choices with respect to changes of the monetary base. For these non-monetary inflation determinants, I closely follow Calderón and Schmidt-Hebbel (2010) and consider the following potential determinants of inflation: Exchange rate regime: Fixed exchange rate regimes are expected to be associated with lower inflation rates due to their disciplinary effect: If the domestic inflation rate persistently exceeds the inflation rate of the currency it is pegged to, a currency crisis may be caused. This effect might be empirically weak since the choice of a fixed exchange rate system is endogenous: A fixed exchange rate regime is often adopted by countries suffering from high inflation rates in order to stabilize their inflation. Since a fixed exchange rate enhances the central bank’s credibility, it may also contribute to lower inflation expectations. Trade openness: Openness reduces the net marginal benefit of surprise inflation. Surprise monetary expansions depreciate the real exchange rate, leading to a negative effect on the terms-of-trade. Since the fraction of foreign goods in domestic consumption increases in the degree of an economy’s openness, an open economy suffers more from a negative terms-of-trade shock. Openness is therefore expected to be associated with lower inflation rates. Capital account openness: The integration of an economy in the world capital market is expected to reduce its inflation rate. An open capital account increases the benefits of low inflation rates (and low inflation variability). First, low inflation decreases the costs of foreign capital and signals sound macroeconomic policies. Second, under an open capital account, domestic agents can more easily substitute domestic currency holdings by foreign currency. Since currency substitution increases if inflation rates are high, central banks might fear a reduction in domestic money demand. Global inflation: Global inflationary tendencies might be an important determinant of domestic inflation (see Section 2). We account for global inflation by the GDP-weighted average inflation rate across countries. Alternatively, we include the oil price to account for higher production costs, which are exogenous from the perspective of small oil-importing countries. Public debt: Governments running persistent fiscal deficits may eventually rely on inflation to finance its debt burden. Inflation has two desired effects: It generates revenues (seigniorage) and lowers the real value of government debt. Public debt is therefore expected to produce inflation. Per capita income: GDP per capita is included as a proxy for economic development and institutional quality. More developed countries are expected to have lower inflation rates. Financial development: Financial development, proxied by domestic credit to the private sector relative to GDP, is expected to be negatively correlated with inflation. First, it proxies the institutional quality of a country. Second, government deficits can more easily be financed by capital markets, reducing the importance of seigniorage income. Third, the financial sector’s distaste for inflation might influence the decision-making in favour of less inflation. Democracy: Our index variable for political regime characteristics controls for the institutional environment. It may also be considered as a proxy for the independence of the central bank.12 Central banks that depend less on political pressures to generate seigniorage income or to foster economic growth, are less likely to generate political business cycles.13

12

Time-series data on central bank independence is not available for our broad sample of countries. For a panel data study showing that monetary policies are more expansionary before elections than in normal times, see Dreher and Vaubel (2009). This effect, however, also holds for independent central banks. 13

130

A. Steiner / North American Journal of Economics and Finance 41 (2017) 112–132

Financial crisis: Periods of financial distress might be characterised by higher inflation rates. Our dummy variable for crisis episodes controls for banking crises, currency crises and sovereign debt crises. High inflation/hyperinflation: Outlier episodes are accounted for by the inclusion of our previously defined dummy variables for high inflation periods and hyperinflations. Inflationary inertia: Current inflation might not be time-independent but related to past inflation rates. This might be due to pricing policies and the formation of inflation expectations. If indexation mechanisms are in place such that prices like wages, tariffs and rents adjust automatically to past inflation rates, the current price level depends on the past inflation rate. Adaptive expectations imply that the expected future inflation rate is the higher, the higher the inflation in the past was. International reserves: We augment this standard set of inflation determinants by the growth rate in international reserves. Central banks’ reserve policies might affect inflation via money supply in addition to the set of non-monetary inflation determinants. We estimate the following fixed effects regression using OLS

^ i;t þ c  X i;t þ ci þ dt þ 2i;t pi;t ¼ aNFA ^ is measured by its average value over the preceding where d is a fixed time effect. To account for the dynamic structure, N FA three years. The results are presented in Table D. Columns 1 and 2 make use of the full sample and control for high- and hyperinflations by the inclusion of dummy variables. Beginning with column 3 the sample is restricted to moderate inflation rates (annual inflation <50%). Column 4 uses an instrumental variable approach, where potentially endogenous variables (trade openness, capital account openness, public debt, GDP per capita and domestic credit) are instrumented by their first lag. The results may be summarised as follows: Fixed exchange rate regimes are associated with lower inflation rates. The effect is significant in four out of nine specifications. In those cases, where trade openness significantly affects inflation, the effect is positive. While this effect is in contrast to our expectations, it confirms the findings of Alfaro (2005) and is theoretically explained by Cooke (2010). Capital account openness substantially lowers inflation. Global inflation and financial crises are correlated with higher inflation rates. In the sample of moderate inflation rates, democratic regimes are associated with lower inflation. The growth in foreign reserves contributes to significantly higher inflation rates in all specifications restricted to moderate inflation. Its economic impact is relatively small: At the mean of the growth rate of reserves averaged over three years (16%), reserves increase inflation by 0.2 percentage points. To account for inflation inertia, we include the lagged level of inflation as an additional determinant. These dynamic specifications are estimated by the Arellano-Bond estimator (column 8) and the system GMM estimator (column 9). Reserves enter by their contemporaneous growth rate in the regression specification; due to the features of the dynamic estimators, only lagged values of reserve growth and, in the case of the GMM estimator, in addition lagged first differences of reserve growth are used for estimation. The Sargan test, which tests the validity of the over identifying restrictions, is in favour of our specification. By and large, previous results are confirmed. Past inflation significantly affects its contemporaneous rate. Determinants of inflation. (1)

(2)

(3)

(4)

(5)

(6)

(7)

Inflation, lagged Fixed exchange 9.1948 1.0268 rate, dummy (0.58) (0.07) Trade openness 0.2058 0.2194 (1.01) (0.75) Capital account 4.9831 15.4802⁄⁄ Openness (0.54) (2.18) Global inflation 0.4156 0.7137 (0.49) (1.17) Public debt 0.6962 0.8434 (1.61) (1.04) GDP per capita 0.0015 0.0029⁄⁄ (0.65) (2.14) Domestic 0.6268 0.5032 credit (0.90) (0.99) Democracy 2.1372⁄ 2.7277⁄⁄ (1.87) (2.07)

0.8996 (1.25) 0.0166 (1.14) 2.4474⁄⁄⁄ (6.34) 0.0628⁄⁄⁄ (3.07) 0.0136 (0.90) 0.0002 (1.61) 0.0388⁄⁄

0.8726⁄ (1.77) 0.0155 (1.49) 2.2565⁄⁄⁄ (11.63) 0.0650⁄⁄⁄ (3.66) 0.0111 (1.48) 0.0002⁄⁄⁄ (3.21) 0.0336⁄⁄⁄

2.1320⁄⁄ (2.49) 0.0313⁄ (1.72) 2.1138⁄⁄⁄ (5.27) 0.2974⁄⁄⁄ (5.98) 0.0395⁄⁄ (2.50) 0.0000 (0.22) 0.0107

(2.53) (3.61) (0.74) 0.2807⁄⁄ 0.2627⁄⁄⁄ 0.0539 (2.57) (5.36) (0.48)

(8)

(9)

0.3674⁄⁄⁄

0.6002⁄⁄⁄

(13.48) 1.4835⁄⁄⁄ (3.05) 0.0853⁄⁄⁄ (15.79) 1.4932⁄⁄⁄ (4.46) 0.0875⁄⁄⁄ (26.69) 0.0195⁄⁄⁄ (2.74) 0.0003⁄⁄⁄ (4.13) 0.0329⁄⁄⁄

(14.92) 0.7555⁄ (1.91) 0.0033 (1.25) 0.5500⁄⁄⁄ (2.83) 0.1026⁄⁄⁄ (7.60) 0.0011 (0.13) 0.0000 (0.96) 0.0184⁄⁄⁄

2.1320⁄⁄ (2.49) 0.0313⁄ (1.72) 2.1138⁄⁄⁄ (5.27) 0.0333⁄⁄⁄ (5.30) 0.0395⁄⁄ (2.50) 0.0000 (0.22) 0.0107

0.4283 (0.53) 0.0401⁄⁄ (2.25) 2.1225⁄⁄⁄ (5.56) 0.0902⁄⁄⁄ (3.91) 0.0027 (0.17) 0.0001 (0.82) 0.0483⁄⁄⁄

(0.74) 0.0539 (0.48)

(3.04) (4.98) 0.3262⁄⁄⁄ 0.0884 (3.37) (1.29)

(3.48) 0.0045 (0.13)

131

A. Steiner / North American Journal of Economics and Finance 41 (2017) 112–132

Appendix D (continued) (1) Financial crisis Foreign 2.2319 reserves, growth (0.55) Hyperinflation 4410⁄⁄⁄ (3.69) Highinflation 99.08⁄⁄⁄ (3.80) Observations 2832 Number of 116 countries R-squared 0.36 Fixed time no effects

(2)

(3)

(4) ⁄⁄⁄

(5) ⁄⁄⁄

(6) ⁄⁄⁄

(7) ⁄⁄⁄

(8) ⁄⁄⁄

(9) ⁄⁄⁄

8.3748 (1.09) 0.1902

3.3720 (4.63) 1.3817⁄⁄⁄

2.8942 (7.67) 1.2682⁄⁄

2.6295 (4.17) 1.0090⁄⁄

2.6295 (4.17) 1.0090⁄⁄

2.3882 (3.26) 3.4812⁄⁄⁄

1.8147 (6.04) 1.0077⁄⁄⁄

1.3451⁄⁄⁄ (3.00) 0.9047⁄⁄⁄

(0.03) 3424⁄⁄⁄ (3.60) 110.80⁄⁄⁄ (3.90) 1734 63

(2.69)

(2.23)

(2.15)

(2.15)

(5.86)

(5.20)

(2.51)

1600 63

1526 63

1600 63

1600 63

1592 63

1527 62

1640 63

0.54 yes

0.26 no

no

0.35 yes

0.35 yes

0.18 no

no

no

Notes: Estimation of columns 1 to 7 by OLS including fixed country effects. Column 4 uses an instrumental variable approach, where potentially endogenous variables are instrumented by their lags. The dynamic specifications are estimated by the Arellano-Bond estimator (column 8) and the system GMM estimator (column 9). t-statistics (in brackets) computed with heteroskedasticity-consistent standard errors. ⁄, ⁄⁄, and ⁄⁄⁄ denote significance at the 10%, 5%, and 1% levels, respectively. Beginning with column 3 the sample is restricted to moderate inflation rates (annual inflation <50%). Column 6 uses oil price inflation instead of world inflation. Column 7 uses inflation calculated by the GDP deflator instead of CPI inflation as dependent variable.

Appendix E. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.najef. 2017.03.007. References Abbas, S. M. A., Belhocine, N., El-Ganainy, A., & Horton, M. (2010). A historical public debt database. IMF Working Paper WP/10/245, Washington, DC. Abdullateef, U., & Waheed, I. (2010). External reserve holdings in Nigeria: Implications for investment, inflation and exchange rate. Journal of Economics and International Finance, 2(9), 183–189. Aizenman, J., & Glick, R. (2009). Sterilization, monetary policy, and global financial integration. Review of International Economics, 17(4), 777–801. Aizenman, J., & Ito, H. (2012). Trilemma policy convergence patterns and output volatility. The North American Journal of Economics and Finance, 23(3), 269–285. Aizenman, J., & Lee, J. (2007). International reserves: Precautionary versus mercantilist views, theory and evidence. Open Economies Review, 18(2), 191–214. Alfaro, L. (2005). Inflation, openness, and exchange-rate regimes: The quest for short-term commitment. Journal of Development Economics, 77(1), 229–249. Assibey-Yeboah, M., & Mohsin, M. (2014). The real effects of inflation in a developing economy with external debt and sovereign risk. The North American Journal of Economics and Finance, 30, 40–55. Barnett, W. A. (1980). Economic monetary aggregates: An application of index number and aggregation theory. Journal of Econometrics, 14, 11–48. Barro, R. J. (1997). Determinants of economic growth. A cross-country empirical study. MIT Press Cambridge Mass. Batini, N., & Nelson, E. (2001). The lag from monetary policy actions to inflation: Friedman revisited. International Finance, 4, 381–400. Belke, A., Beckmann, J., & Czudaj, R. (2014a). The importance of global shocks for national policymakers – Rising challenges for sustainable monetary policies. The World Economy, 37(8), 1101–1127. Belke, A., Beckmann, J., & Czudaj, R. (2014b). Does global liquidity drive commodity prices? Journal of Banking and Finance, 48 (C), 224–234. Belke, A., Bordon, I., & Volz, U. (2012). Effects of global liquidity on commodity and food prices. World Development, 44(C), 31–43. Belke, A., Orth, W., & Setzer, R. (2010). Liquidity and the dynamic pattern of asset price adjustment: A global view. Journal of Banking and Finance, 34(8), 1933–1945. Belke, A., & Rees, A. (2014). Globalisation and monetary policy – A FAVAR analysis for the G7 and the Euro Area. North American Journal of Economics and Finance, 29(C), 306–321. Benati, L. (2009). Long run evidence on money growth and inflation. Working Paper Series 1027. European Central Bank. Berger, H., & Harjes, T. (2009). Does global liquidity matter for monetary policy in the Euro area? International Finance, 12(1), 33–55. Calderón, C., & Schmidt-Hebbel, K. (2010). What drives inflation in the world? In R. Fry, C. Jones, & C. Kent (Eds.), Inflation in an era of relative price shocks. Reserve Bank of Australia. Calvo, G. A. (1991). The perils of sterilization. IMF Staff Papers, 38(4), 921–926. Cheng, G. (2015). A growth perspective on foreign reserve accumulation. Macroeconomic Dynamics, 19, 1358–1379. Chinn, M. D., & Ito, H. (2006). What matters for financial development? Capital controls, institutions, and interactions. Journal of Development Economics, 81 (1), 163–192. Chitßu, L. (2016). Reserve accumulation, inflation and moral hazard: Evidence from a natural experiment. Working Paper Series 1880. European Central Bank. Ciccarelli, M., & Mojon, B. (2010). Global inflation. Review of Economics and Statistics, 92(3), 524–535. Cooke, D. (2010). Openness and inflation. Journal of Money, Credit and Banking, 42, 267–287. De Grauwe, P., & Polan, M. (2005). Is inflation always and everywhere a monetary phenomenon? Scandinavian Journal of Economics, 107(2), 239–259. Dornbusch, R., & Fischer, S. (1993). Moderate inflation. World Bank Economic Review, 7, 1–44. Dreher, A., & Vaubel, R. (2009). Foreign exchange intervention and the political business cycle: A panel data analysis. Journal of International Money and Finance, 28(5), 755–775. ECB (2006). The accumulation of foreign reserves. By an International Relations Committee Task Force, Occasional Paper No. 43 (February). Eichengreen, B., El-Erian, M., Fraga, A., Ito, T., Pisani-Ferry, J., Prasad, E., ... Yu, Y. (2011). Rethinking central banking. Washington: Brookings Institution. Estrella, A., & Mishkin, F. S. (1997). Is there a role for monetary aggregates in the conduct of monetary policy? Journal of Monetary Economics, 40, 279–304.

132

A. Steiner / North American Journal of Economics and Finance 41 (2017) 112–132

Ghosh, A. R., Gulde, A.-M., & Wolf, H. C. (2002). Exchange rate regimes: Choices and consequences. Cambridge, Mass: The MIT Press. Ghosh, A. R., Ostry, J. D., & Tsangarides, C. (2012). Shifting motives: explaining the buildup in official reserves in emerging markets since the 1980s. IMF Working Paper No. 12/34. Ghosh, A. R., & Philips, S. (1998). Warning: Inflation may be harmful to your growth. IMF Staff Papers, 45, 672–713. Hauner, D. (2006). A fiscal price tag for international reserves. International Finance, 9, 169–195. Heller, R. (1976). International reserves and world-wide inflation. IMF Staff Papers, 23(1), 61–87. Heller, R. (1979). Further evidence on the relationship between international reserves and world inflation. In M. J. Boskin (Ed.), Economics and human welfare: essays in honor of Tibor Scitovsky. New York. Hendrickson, J. R. (2014). Redundancy or mismeasurement? A reappraisal of money. Macroeconomic Dynamics, 18, 1437–1465. Henriksen, E., Kydland, F. E., & Šustek, R. (2013). Globally correlated nominal fluctuations. Journal of Monetary Economics, 60(6), 613–631. Ilzetzki, E. O., Reinhart, C. M., & Rogoff, K. (2010). Exchange rate arrangements into the 21st century: Will the anchor currency hold? mimeo. IMF (2010a). Reserve accumulation and international monetary stability. Policy and Review Department: Paper prepared by the Strategy. IMF (2010b). Global liquidity expansion: Effects on ‘‘receiving”economies and policy response options. In Global financial stability report. Washington: World Economic and Financial Surveys. Ireland, P. N. (2014). The classical theory of inflation and its uses today. Paper prepared for the November 2014 SOMC meeting. mimeo. Khan, M. S. (1979). Inflation and international reserves: A time-series analysis. IMF Staff Papers, 26, 699–724. Lavigne, R. (2008) Sterilized intervention in emerging-market economies: trends, costs, and risks. Bank of Canada discussion paper 2008-4 Leeper, E. M., & Nason, J. M. (2014). Bringing Financial Stability into Monetary Policy. Report Prepared for Sveriges Riksbank. CAMA working paper 72/2014. Levy Yeyati, E. (2008). The cost of reserves. Economic Letters, 100(1), 39–42. Lin, M.-Y., & Wang, J.-S. (2009). Foreign exchange reserves and inflation: An empirical study of five East Asian economies. Empirical Economics Letters, 8(5), 487–493. Lucas, R. E. (1980). Two illustrations of the quantity theory of money. American Economic Review, 70(5), 1005–1014. Marshall, M. G., & Jaggers, K. (2013). Polity IV project: political regime characteristics and transitions, 1800-2010. Available at URL: http://www. systemicpeace.org/polity/polity4.htm. Mumtaz, H., & Surico, P. (2012). Evolving international inflation dynamics: World and country-specific factors. Journal of the European Economic Association, 10, 716–734. Ndikumana, L., & Elhiraika, A. (2007). Reserves accumulation in African countries: Sources, motivations, and effects. UMASS Amherst Economics Working Papers 2007–12. Neely, C. J., & Rapach, D. E. (2011). International comovements in inflation rates and country characteristics. Journal of International Money and Finance, 30(7), 1471–1490. Nelson, E. (2008). Why money growth determines inflation in the long run: answering the Woodford critique. Journal of Money, Credit and Banking, 40, 1791–1814. Niehans, J. (1976). How to fill an empty shell. American Economic Review, 66(2), 177–183. Obstfeld, M., Shambaugh, J. C., & Taylor, A. M. (2010). Financial stability, the trilemma, and international reserves. American Economic Journal: Macroeconomics, 2(2), 57–94. Ouyang, A. Y., Rajan, R. S., & Willett, T. (2008). Managing the monetary consequences of reserve accumulation in emerging Asia. Global Economic Review, 37 (2), 171–199. Ouyang, A. Y., Rajan, R. S., & Willett, T. (2010). China as a reserve sink: The evidence from offset and sterilization coefficients. Journal of International Money and Finance, 29, 951–972. Pesaran, M. H., & Smith, R. (1995). Estimating long-run relationships from dynamic heterogeneous panels. Journal of Econometrics, 68(1), 79–113. Rabin, A., & Pratt, L. (1981). A note on Heller’s use of regression analysis. IMF Staff Papers, 28(1), 225–229. Reinhart, C. M., & Reinhart, V. R. (2008). Capital inflows and reserve accumulation: The recent evidence. NBER Working Paper 13842. Reinhart, C. M., & Rogoff, K. S. (2011). From financial crash to debt crisis. American Economic Review, 101(5), 1676–1706. Rodrik, D. (2006). The social cost of foreign exchange reserves. International Economic Journal, 20(3), 253–266. Rüffer, R., & Stracca, L. (2006). What is global excess liquidity, and does it matter? ECB Working Paper Series No. 696. Sargent, T. J., & Wallace, N. (1973). Rational expectations and the dynamics of hyperinflation. International Economic Review, 14(2), 328–350. Shrestha, P. K. (2016). Macroeconomic impact of international reserves: Empirical evidence from South Asia. NRB Working Paper 32/2016. Nepal Rastra Bank, Research Department. Shrestha, P. K., & Semmler, W. (2015). Monetary policy and international reserves: Empirical evidence from East Asian countries. International Journal of Finance & Economics, 20, 191–205. Smets, F. (2014). Financial stability and monetary policy: How closely interlinked? International Journal of Central Banking, 10(2), 263–300. Spencer, P., & Liu, Z. (2010). An open-economy macro-finance model of international interdependence: The OECD, US and the UK. Journal of Banking and Finance, 34(3), 667–680. Thornton, D. L. (2014). Monetary policy: Why money matters (and interest rates don’t). Journal of Macroeconomics, 40, 202–213. Woodford, M. (2008). How important is money in the conduct of monetary policy? Journal of Money, Credit and Banking, 40, 1561–1598. Zhang, C. (2013). Monetary dynamics of inflation in China. The World Economy, 36(6), 737–760.