Revisiting the quantity theory of money in Euro Area: the case of Greece.

• Author: Ongan, Serdar

1. Introduction

   The history of the The Quantity Theory of Money (QTM) goes back to at least the 16th century. The French philosopher Jean Bodin (1568) first introduced his monetary theory of the price inflation that occurred in Western Europe caused by new monetary metal inflows from South America (Humphrey, 1974). Following Bodin this structural relationship between quantity of money and inflation has attracted the substantial attention of many economists, such as John Locke (1692), David Hume (1752), Milton Friedman (1968), Karl Brunner and Allen Meltzer (1963), and Ludwig von Mises (1912) with many revisions, as well as numerous cases of further elaboration, and extension between the 16th and the 20th centuries. Apart from these authors' contributions, which are beyond the scope of this study, the QTM, as a cornerstone of monetarism, is built on an equation, namely the equation of exchange, developed by Irving Fisher (1911). According to this equation, presented below, money stock multiplied by the velocity of money equals the nominal GDP.

   M x V = P x R = Y (1)

   In this equation, M is money stock, V is velocity of money, P is the average price level, and R is the real income (PxR = Y denotes nominal income). In the purest and shortest form of the equation, both V and R are expected to be constant, at least in the short-run. Therefore, the QTM in this equation indicates that change in money stock (M) leads to one-to-one (unitary) proportional change in price level (P) in the long-run. Friedman (1963) postulates this relation with his famous dictum "Inflation is always and everywhere a monetary phenomenon". He succinctly and implicitly summarizes that there is a linear (positive) relation between money stock and prices (inflation) (Hetzel, 2007, p.16). According to this linear relation, while rises in money stock (exogenously by central banks) lead to increases in price level, drops lead to decreases. However, this relation (in practice), may be nonlinear (asymmetric). This means that while rises in money stock may lead to decreases in price level, drops may lead to increases. Another possibility is that, while rises in money stock may lead to increases in price level, drops may have no impacts in prices or vice versa. Although there are many studies which have investigated the relation between money stock and inflation (Friedman, 1963; Friedman and Swartz, 1963; King, 2002; Benati, 2009 among others), there is limited number of studies which investigate the asymmetry in pass-through from money stock to inflation (Bell and Mankiw, 1994; Crowder, 1998; Weise, 1999; Karras and Stokes, 1999; Senda, 2001; Amisano and Colavecchioz, 2013; Cooray and Khraief, 2018; Olayiwola and Ogun, 2019).

   The rationale of using an asymmetric (nonlinear) approach in our model is that rising uncertainties in economies and asymmetric information problem can easily cause potential asymmetric (nonlinear) behaviors/results in financial markets such as adverse selections, moral hazards, incomplete markets and thereby market failures. Additionally, adaptive expectations, changing money demand motives (transaction, precautionary and speculative) and financial crises may increase these asymmetric (nonlinear) relations also between macroeconomic variables such as money stock and inflation. This means that all economic actors such as barrowers, lenders, financial intermediaries, and central banks may exhibit aymmetric behaviours potentially. Hence, all these may require also applying nonlinear (asymmetric) approaches in empirical models. The Liquidity Trap by Keynes (1936) is a good example for this potential nonlinear (asymmetric) relation between money stock and inflation. If increases in money demand are proportionally equal to increases in money stock, inflation may instead remain stable or increases in money demand are higher than increases in money stock, inflation may fall.

   Therefore, this study revisits the QTM from this perspective of potentially asymmetric (nonlinear) relations and reexamines this theory for Greece. To this aim, for the first time, the nonlinear ARDL (Auto Regressive Distributed Lag) model is applied for testing the QTM for a specific country. This methodological approach makes this study different from previous empirical studies, which test the QTM on the assumption of linear (symmetric) relations for Greece or any other country. Few studies empirically examine the QTM for Greece. Karfakis (2002) applies the unit root and the ARDL approach to cointegration and tests two monetarist hypotheses, i.e., the predictability of income velocity of money and the proportionality between money stock and nominal income (or, prices). The author finds the validity of the QTM for this country. Ozmen (2003), in response to Karfakis (2002), applies the ARDL bounds and Johansen procedures and does not find the validity of the QTM for Greece. Karfakis (2004) again, in response to Ozmen (2003) applies the ARDL approach with the maximum lag at six in the VAR (vector autoregressive) system and re-affirms the exogeneity of money supporting the QTM for Greece.

   Besides its new methodology, another departure point for this study is the sample period used and analyzed. Contrary to the studies mentioned above, this study tests the QTM for the post-period of Greece's adoption (after March 2002) of the Euro. This adoption, as a game changer, required redefinition of the monetary aggregates for all euro area (1) countries, which caused some uncertainties concerning the real amounts of monetary aggregates in these countries. If it is because of these concerns scholars might have intentionally avoided testing the QTM for euro area countries including Greece. However, it is believed that the QTM may be tested for Greece (or for any other euro area country) by using newly defined monetary aggregates of this country, which is referred to as the "Greek contribution (2)" to the euro area aggregates. The Greek contribution is calculated in different manners and equals to: (i) the deposits held by Greek and other euro area countries' residents in Greek monetary financial institutes (MFIs); (ii) the banknotes put into circulation by the Bank of Greece (BoG); (iii) dept securities issued by Greek MFIs minus debt securities issued by all euro area MFIs. Therefore, the empirical results of this study should be considered-interpreted on the assumption that the series of the Greek M1 and M2, which were calculated on the basis of the...