Abstract
This Selected Issues paper for euro area policies analyzes the product market regulation and benefits of wage moderation. The paper identifies structural shifts in the relationship between wages and unemployment rates—a “wage curve”—in 20 industrial countries. It reviews euro area and cross-country developments in labor costs and their bivariate relationship with unemployment rates and business GDP. The paper also examines aspects of the European Central Bank’s monetary analysis, within the context of their overall two-pillar policy framework, and issues surrounding its use.
II. Declining Money Velocity in the Euro Area: Implications for the ECB’s Monetary Analysis19
A. Introduction
30. This chapter examines aspects of the ECB’s monetary analysis, within the context of their overall two-pillar policy framework, and issues surrounding its use. At issue is not whether euro-area inflation is a monetary phenomenon; clearly, at some horizon, money matters. The policy challenge for central banks is to operationalize this in a robust way. In the ECB’s case, a key question is how monetary aggregates can serve as reliable input within it’s monetary policy strategy that aims to achieve price stability over the “medium term.”20 Within the central bank’s two-pillar framework, the monetary analysis should provide, in the first instance, a robust assessment of liquidity conditions, and, ultimately, an effective “cross-check” on inflationary pressures over this horizon. This paper illustrates some empirical dimensions of these challenges by examining the following questions:
What is the nature of the relationship between money and inflation in the euro area over different horizons, particularly over the medium term?
What measurement and conceptual issues affect the assessment of liquidity conditions, particularly in light of the (variable) trend decline in area-wide velocity?
To what extent do macroeconomic versus structural factors help explain the behavior of velocity? What bearing does this have for assessing liquidity conditions, and, ultimately, for the “cross-check” on inflationary risks from the monetary analysis?
Answers to these questions are instrumental in helping delineate what operational role money should play—within the ECB’s policy framework—in order to offer a reliable guide to effective monetary policy decision making and communication.
B. Background
31. The ECB’s monetary policy framework—and the role ascribed to money—has evolved since its inception. Originally, the ECB’s monetary policy framework was elaborated as a two-pillar strategy, with money occupying the preeminent role as the “first pillar” in the analysis of medium-term risks to price stability. Figure II.1 shows a timeline of key events and the evolution of money’s role within the ECB’s policy framework over the past six years. Following the clarification of its policy strategy in May 2003, the ECB moved away from its “pillar” designation and recast the role of money in its inflation assessment. “Monetary analysis,” in the ECB’s parlance, was to be used to provide a medium- to long-term perspective, to “cross-check” the assessment of short- and medium-term risks to price stability obtained from its “economic analysis.” Many outside observers have interpreted these changes as deemphasizing the role of monetary aggregates in the ECB’s policy framework.21
ECB Timeline and the Designation of Money
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
ECB Timeline and the Designation of Money
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
ECB Timeline and the Designation of Money
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
32. However, the ECB’s monetary analysis has figured prominently in its recent concerns and communications. At its Fall 2004 meetings, the Governing Council adopted discernibly “hawkish” language with respect to the balance of risks to price stability, emphasizing the need for “strong vigilance.”22 Faced with a recovery losing steam, a strengthening currency, and inflation already at low levels (albeit slightly above 2 percent), why did the ECB strike such a strident note on inflation risk?23 With policy interest rates on hold at historically low levels for nearly two years, the ECB’s concerns derive, in large part, from its monetary analysis or “cross check” with money. Concerns center around a liquidity “overhang” and uncertainty about how this would unwind once the recovery gathers momentum. More concretely, broad money’s sustained growth in excess of its reference value has accumulated into a sizable “real money gap.” Accordingly, the ECB assesses that “there remains substantially more liquidity in the euro area than is needed to finance non-inflationary growth.” Concomitant concerns that low interest rates were fueling asset price inflation—i.e., a boom in house prices in several area real estate markets—round out the cautionary signals from the cross-check and the need for “continued vigilance” on the part of the ECB.24
33. But recent liquidity developments may pose an “embarrassment of riches”with respect to inflation. On some level, the fact that liquidity has grown rapidly for several years, while underlying inflationary pressures have remained well contained, if not subdued, raises broader questions about the operational robustness of the monetary analysis and its cross check on medium-term price pressures. In response to this challenge, ECB staff analysis has examined complicating issues regarding estimates of excess liquidity and its potential inflationary impact—citing the role of portfolio shifts and risk aversion in the wake of the asset price boom and bust in 2000–01.25 In light of these and other uncertainties, however, a lingering question remains regarding the appropriate level of concern one should infer from recent monetary developments.
34. The paper is organized as follows. Section C recounts the numerical basis for the ECB’s reference value and estimates of “excess” liquidity and the real money gap. Section D briefly revisits the nexus between money and inflation in the euro area. Section E presents (univariate) structural break analysis of area-wide velocity and highlights measurement issues for liquidity given velocity’s variable decline. Section F broadens the discussion to a multivariate context and explores possible factors behind changing liquidity preference in the euro area as it pertains to the robustness of the cross-check with money. Section G concludes.
C. Mind the Gap
35. Money growth has shown little affinity for the reference value since EMU, suggesting the build-up of a“liquidity overhang” in recent years. Signaling the prominent role of money, the numerical face of the ECB’s monetary analysis is the reference value—i.e., 4½ percent for annual M3 growth—defined more precisely than its core inflation objective of “below but close to 2 percent.”26 Actual growth in broad money, however, has persistently exceeded the 4½ benchmark since 2001, adding up to ample liquidity. To illustrate this, the real money gap is shown in the Figure II.2, based on observed growth in M3 less that in consumer prices–relative to their respective reference value or policy objective (see below)—and accumulated assuming an initial gap of zero in January 1999 when the euro was introduced.
Monetary Developments
(In percent)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB; Eurostat and staff calculations.Monetary Developments
(In percent)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB; Eurostat and staff calculations.Monetary Developments
(In percent)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB; Eurostat and staff calculations.36. To better understand the quantitative basis for money’s reference value, the quantity equation furnishes the basic analytical framework. The relationship asserts that the product of money with its turnover rate or velocity of circulation equals the value of all nominal transactions (proxied by nominal GDP); in log differences, the equation can be expressed as follows:
Assuming that the medium-term values for annual inflation π and potential output growth are 1 ¾ percent and 2 percent respectively, equation (1) suggests that the ECB’s reference value for money growth of 4½ percent corresponds to an equilibrium decline in velocity of -¾ percent. Consistent with a reasonable range for potential output growth (1¾-2¼ percent), the ECB’s acknowledged range for the annual change in velocity around this mid-point is -½ to -1 percent.27 The validity of this latter assumption is explored in Section E.
37. Estimates of excess liquidity can also be understood in terms of “gaps,”including for velocity. Note that the quantity relationship can also be written in “gap” form for money, prices, output, and velocity—expressed as deviations from their medium-term (equilibrium) values (denoted with a bar ¯):
Equation (2) shows that the real money gap must equal the difference between the output and velocity gaps. When the economy is cyclically strong (i.e., positive output gap) or the velocity of circulation is depressed (i.e., negative velocity gap), real liquidity should be relatively high. Correspondingly, if money growth has been relatively brisk but has not passed through to higher inflation (or output), velocity must have declined. In terms of numbers, the area’s “excess” liquidity can be illustrated in velocity space as follows. Assuming a zero velocity gap in 1999, the ECB’s assumed range for trend velocity is drawn in Figure II.3.28 The considerable negative deviation in velocity (i.e., between 7–10 percent in mid-2004) from the ECB’s reference range that emerges is consistent with the sizeable positive real money gap shown in Figure II.2 adjusted for output gaps.
Euro Area M3 Velocity, 1980:Q1–2004:Q3
(In logarithms)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB and staff estimates.Euro Area M3 Velocity, 1980:Q1–2004:Q3
(In logarithms)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB and staff estimates.Euro Area M3 Velocity, 1980:Q1–2004:Q3
(In logarithms)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB and staff estimates.D. Horizon Matters
38. A useful policy role for money—in particular, M3—depends upon its role in determining inflation and over what horizon, which are ultimately empirical issues. The choice of M3 as the “right” measure of money is predicated on two key assumptions: (1) the stability of the money demand relationship, and (2) its leading indicator properties for medium-term price developments. The general validity of both assumptions are discussed below. From a policy standpoint, however, it should be noted that even if both conditions were met, whether this justifies money’s special role as a separate pillar in the ECB’s policy framework is a separate issue that is subject of an ongoing, vigorous debate. See Box II.1.
39. Most studies find that the link between money growth and inflation is robust at longer horizons. Using euro-area data from 1980-2004, Neumann and Greiber (2004), for example, find a one-to-one relationship between “core” money growth—defined as long-run movements—and inflation. Monetary cycles less than 8 years, however, have no predictive power. Jaeger (2003) finds the same horizon cut-off for various industrial countries over the past 40 years. A contrary view, however, is offered by Nicoletti-Altimari (2001), arguing that euro-area money and credit aggregates can forecast inflation at shorter horizons. The information content of money is especially useful at horizons beyond 1½ years, and performs best at the longest horizon considered there (3 years). Also, simple monetary and credit aggregates outperform real money gap measures.29
A Separate Pillar for Money?
Defenders of the separate money pillar put forward a number of justifications:
Money can act as a timely proxy. Masuch and others (2003) maintain that since GDP and output gaps are notoriously difficult to identify in real time, monetary indicators can perform a useful function. Indeed, Orpanides (2003) argues that the 1970s inflationary bias in the United States reflected exaggerated beliefs regarding the output gap, given the failure to recognize the productivity slowdown. But Gerlach (2004), for example, argues there is little reason to think that money growth can provide as much information about the output gap as output itself (or its direct proxies).
Money plays a distinct role in assessing financial developments. Masuch and others (2003) point to money’s signaling role in the build-up of financial imbalances and asset price bubbles. Chapter III revisits these issues. Jaeger (2003) argues that having this distinct role may provide some insurance against the emergence of these financial dynamics. Adherents of this viewpoint often argue that an appropriate role for money could have lessened the impact of the Great Depression and the Japanese crash in the late 1980s. But others challenge the notion that this calls for a separate pillar and claim that policymakers should look at more direct measures of financial distress (Galí and others, 2004).
Money serves as a nominal anchor. Jaeger (2003) suggests that the anchor role is rooted in psychology, based on some European countries’ postwar experiences with (hyper-) inflation. Masuch and others (2003) argue that money can help avoid destabilizing paths for inflation expectations and potential problems of indeterminacy associated with “moneyless” policy rules (e.g., Taylor rule), identified by Benhabib et al (2001) and others; McCallum (2002), however, argues that the “expectations trap” issues are of little practical importance. Also, Galí and others (2004) counter by arguing that problems arising from self-fulfilling expectations are not confined to policy rules that exclude money nor is including money necessary to solve the problem.
Critics have argued that the two-pillar system creates manifold communication problems. After the ECB policy strategy clarification in 2003, some have complained that the precise role for its monetary analysis remains confusing, especially since M3 growth has consistently outpaced the reference value. Many have taken this to argue that the ECB does not directly take money growth into consideration when setting interest rates, relying more on indicators of economic activity, suggesting a mismatch between “words and deeds” (Gerlach, 2004). Even supportive studies—finding a role for monetary factors in setting policy rates—agree that variables like the real money gap are more pertinent than raw M3 growth rate (Carstensen, 2003). Others have criticized the current situation as one of monetary policy “immobility” but communications “volatility” (in tone), reflecting an ECB “caught between two pillars” yielding conflicting signals (Gros, et al, 2005).
40. A simple glance at the euro-area data suggests that the link between money and inflation is more apparent at lower frequencies. Figure II.4 compares (non-overlapping) changes in (log) prices—measured either by the GDP deflator or the HICP index—with changes in (log) M3 in excess of (log) real GDP at quarterly, annual, 5-year, and 10-year horizons. Although correlations (denoted by ρ in the figure) are low at shorter horizons, the correlation between inflation and “excess” money growth is near unity at five years and ten years. Notably, however, “too much money chasing too few goods” has tended to over predict goods price inflation in the euro area. Namely, money growth in excess of output growth has translated into less than proportional inflation (i.e., observations lying below the 45-degree line), reflecting velocity’s trend decline.30 A notable outlier in this direction is the 5-year period under EMU from 2000–04 (see Figure II.4, Panel 3). Hence, ascertaining the likely inflationary impact of monetary developments requires a more explicit accounting for velocity’s trend behavior and the factors underpinning it.
Euro Area Money and Inflation Correlations
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB and staff estimates.Euro Area Money and Inflation Correlations
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB and staff estimates.Euro Area Money and Inflation Correlations
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB and staff estimates.E. Measuring Velocity Trends
41. The usefulness of the cross-check provided by the monetary analysis hinges on its reliability in assessing liquidity conditions—which faces challenges stemming from both measurement and conceptual issues. While money growth may adequately explain inflation over the longer-term, robust assessment of medium-term risks to price stability is faced with additional challenges associated with changing velocity or liquidity preference, including those possibly induced by the advent of the single currency. Three areas—two narrow and one broad—are pertinent:
First, analysis of area-wide money demand is confronted by the short sample of the most relevant time period—i.e., that provided by monetary union itself. Historical lessons are inherently difficult to draw from the earlier period given that different monetary regimes across member states existed prior to the start of EMU.
Second, reliably measuring liquidity conditions in the context of changing trends in velocity becomes more difficult. This challenge is most apparent for the numerical focal point of the monetary analysis—namely, M3’s reference value.
Finally, and more broadly speaking, robustness of the monetary analysis and its cross-check for underlying inflation will crucially depend on determining the economic forces (i.e., transitory or structural) driving velocity changes as developments unfold. For the ECB, it must ascertain in real time the operational implications of these dynamics. These issues are discussed further in the current and penultimate sections, respectively.
42. Historical analysis of area-wide velocity is immediately confronted with a fundamental uncertainty regarding data prior to the formation of monetary union. The most relevant period to examine money demand under EMU covers only about six years. In addition to the inherent uncertainty in relating economic behavior before and after monetary union, a more basic data uncertainty arises on how best to combine past, diverse national experiences—i.e., under previous monetary and exchange rate regimes—to inform analysis about EMU.31 To make this concrete, consider several alternative series for M3 velocity—all constructed using ECB data and shown in Figure II.5. The series labeled “ECB”, shown earlier and used primarily for the analysis, is the central bank’s preferred velocity series. Its construction is based on aggregated national data for nominal GDP and M3 at irrevocable fixed exchange rates, announced on December 31, 1998 (and as from January 1, 2001, in the case of Greece).32 A shorter series (labeled “PPP”) combines national outputs and money stocks at purchasing power parities, following the methodology of the ECB’s own Area Wide Model (AWM).33 Conceptually, in the aggregation, the former construction emphasizes the nominal relevance of each member state’s economy (anchored by its 1998 fixed nominal conversion rate), while the latter places weight on the real relevance of each economy (as measured by its PPP benchmark). Yet a third series (labeled “VSA”) would join aspects of both approaches and use a variable-specific aggregation, based on the concept “closest” to the variable (e.g., nominal-based or real-based weights), combining monies at irrevocable exchange rates and outputs at PPP. Looking at the figure, the three series all show a secular decline in velocity but do not yield identical pictures. Unfortunately, a clear choice is not obvious since both real and nominal factors are relevant for liquidity conditions and demands. Detailed analysis based on the VSA series, by virtue of its longer time span—but similar to the results obtaining with the shorter PPP series, is relegated to the appendix. But key differences with the ECB series are noted in the text. It should also be noted that all three methodologies use fixed weights and tend to produce less volatile turnover rates (i.e., more predisposed toward stable velocity) than (say) one based on market exchange rates.
Alternative Historical Velocity Series, 1970:Q1–2004:Q2
(In logarithms)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB and staff estimates.Alternative Historical Velocity Series, 1970:Q1–2004:Q2
(In logarithms)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB and staff estimates.Alternative Historical Velocity Series, 1970:Q1–2004:Q2
(In logarithms)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB and staff estimates.43. Regarding numerical assessments, the economic relevance of the real money gap relies on the sensibility of its reference values, including for velocity. The largest sources of uncertainty in the analysis are on the right-hand side of equation (1). On the left hand side, the medium-term value for inflation is a normative concept—i.e., an externally-given objective of policy; here, the only issue is ascribing a numerical value or “thick point” for the ECB’s inflation objective. Also, supply-side analyses offer independent estimates of potential growth and the output gap. This turns the focus to the other side of the equation and the reliability of benchmarks for money and velocity, which are inextricably intertwined. First, velocity (and its gap) cannot be observed independently from (and thus embodies) developments in money, output, and prices. Second, given the inflation objective and potential growth, the economic relevance of the reference values for money and velocity necessarily go hand-in-hand.
44. Contrary to various expectations, velocity in the euro area has continued its secular decline. Financial innovation and numerous technological advances creating greater access and viable alternatives to money would suggest that its turnover rate should increase over time. Moreover, with the euro’s introduction, many had expected that a wider menu of financial instruments (e.g., deeper securities markets) would reduce the role of money (and credit) in the euro area economy.34 However, the history of money velocity stands in direct contradiction to these predictions—and not just since EMU, but over the past quarter century.35 The velocity of circulation (as evident in Figure II.3 and II.5) has experienced a long-run decline—i.e., money is being held longer and has been rising relative to GDP.36 Based on ECB quarterly data from 1980 to 2004 for M3 and nominal GDP, a simple linear trend model for the quarterly series yields the following estimates:
Both the constant and trend coefficient are significant at the 1 percent level. The trend coefficient, obtained with quarterly data, suggests (on an annualized basis) a decline in velocity of roughly -1 percent per year—not far from the ECB’s (mid-point) assumption. However, several econometric problems are evident. First, standard unit root (i.e., ADF) and stationarity (i.e., KPSS) tests both suggest that the linear trend model fares poorly compared to a non-stationary model; see Table II.A1. In economic terms, this highlights the fact that velocity deviations from trend are extremely persistent, leaving open the issue of whether velocity shocks are permanent or whether the trend is misspecified. Stability tests—valid under the assumption of a trend stationary model—reject parameter constancy for the trend; see Table II.A2.37
45. Structural change analysis of area-wide velocity indicates at least one trend break in the series. Visually, in Figure II.3, the declining trend appears to have steepened more recently. More formally, sequential regressions testing for a one-time trend break in velocity across all permissible break dates—based on the Quandt-Andrews maximal F-test—suggest a significant break around 2001:Q4; see Figure II.6.38 Using a battery of other structural change tests based on various methodologies, other (single) break dates obtain in the narrow range from 2000:Q2 to 2001:Q4; see Table II.A2. For example, Perron’s (1997) unit root test against a stationary alternative with a breaking trend (with endogenous or unknown breakpoint) selects a similar break date, based on maximizing the possibility of rejecting a unit root.39 Interestingly, but perhaps not surprisingly (given low statistical power in finite samples), this latter test still does not strongly reject a unit root—indicating that deviations around the breaking trend are still fairly persistent and leaving open the possibility of multiple breaks.
Euro Area Velocity, 1980:Q1–2004:Q3
(In logarithms unless noted otherwise)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB and staff estimates.1 Quandt-Andrews F-statistic (in levels).Euro Area Velocity, 1980:Q1–2004:Q3
(In logarithms unless noted otherwise)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB and staff estimates.1 Quandt-Andrews F-statistic (in levels).Euro Area Velocity, 1980:Q1–2004:Q3
(In logarithms unless noted otherwise)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB and staff estimates.1 Quandt-Andrews F-statistic (in levels).46. Further tests indicate possible multiple breaks in velocity, particularly if the longer series is considered. Formal tests for multiple structural breaks—using both sequential and the simultaneous methods—suggest additional breaks points in the ECB velocity series around 1994; see Table II.A2.40 The longer VSA series for velocity also reveals several potential break points, with a dominant break region around 1987; see Table II.A3.41 In that case, additional break tests corroborate the most recent break region between end-2000 to end-2001 found in the ECB series (see Figure II.7).
Euro Area M3 Velocity, 1980:Q1–2004:Q3
(In logarithms)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB and staff estimates.Euro Area M3 Velocity, 1980:Q1–2004:Q3
(In logarithms)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB and staff estimates.Euro Area M3 Velocity, 1980:Q1–2004:Q3
(In logarithms)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB and staff estimates.47. Working with a single structural break yields a very different estimate of (post-break) trend velocity from the linear model. Velocity, allowing for a non-linear trend function drawn in Figure II.6, can be represented by:
Including a trend break, the fit of the equation improves and parameter instability is attenuated (see Table II.A2). In breaking-trend form, the implied annual rate of decline in velocity steepens from -1 percent prior to 2001 to -2½ percent thereafter, exceeding (in absolute value) the upper end of the ECB’s range between -0.5 and -1.0 percent for trend velocity growth. Table II.1 compares velocity trends under both models with the ECB’s assumed range for velocity.
Euro Area M3 Velocity Trends, 1980:Q1-2004:Q3
(Annualized rate; in percent)
(Annualized) standard errors in parentheses, calculated using delta method.
48. Results regarding velocity’s variable decline are robust to the assumption of trend stationarity versus difference stationarity. Unlike the trend model, the growth model—i.e., working in first differences of (log) velocity—does not suffer from problems associated with unit roots (see again Table II.A1). But very similar implications for velocity’s trend obtain from time-series estimates based on velocity growth; see Table II.A4.42 Specifically, the growth model also shows signs of parameter instability, namely in the constant term (analogous to the slope coefficient in the trend model).43 Moreover, allowing for a shift in mean—i.e., in the drift—yields similar estimates for velocity’s post-break decline and very similar location(s) of the breakdate(s); see Table II.A4.
49. Estimates of “excess” liquidity are sensitive to underlying assumptions regarding the behavior of trend velocity. The implications of velocity’s breaking-trend (or mean-shift in growth) for the real money gap can be illustrated as follows. With a single structural break, the relevant reference value for annual M3 growth would rise (post-break) to around 5¾ or 6 percent rather than 4½ percent, ceteris paribus.44 To put these numbers in some perspective, average M3 growth has been 6½ percent since the start of monetary union; see Figure II.8. Using the same initial assumptions—i.e., zero gaps for velocity, output, and real money at the start of EMU—the accumulated real money gap in 2004:Q2 would be 6½-7½ percentage points lower than in Figure II.2. Relaxing these starting assumptions (i.e., zero gaps in 1999), one could further use the level implications of the breaking-trend model. Here, a very small, positive velocity gap emerges, implying a negative real money gap, albeit close to zero.
Distribution of Money Growth, 1999:Q1-2004:Q2
(In percent)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB and staff calculations.Distribution of Money Growth, 1999:Q1-2004:Q2
(In percent)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB and staff calculations.Distribution of Money Growth, 1999:Q1-2004:Q2
(In percent)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB and staff calculations.F. Explaining Changing Liquidity Preference
50. While the structural break analysis highlights potential instability in velocity, it cannot explain the nature and causes of changing liquidity preference—instrumental for determining money’s policy role. The univariate analysis cannot provide an economic explanation for the “trend” breaks that have been identified. Conceptually, these structural breaks may even be compatible with multivariate analysis showing a stable long-run demand for money—depending on its economic determinants (and their behavior). Seen in this broader context, area-wide velocity has declined at a time of low inflation, interest rates, and asset returns that have reduced the opportunity cost of holding money. These macroeconomic factors could explain rising money demand in an otherwise predictable fashion. A number of ECB studies indeed find a stable long-run money demand function, although using sample periods that typically end in 1999 (or earlier).45 Taking a more critical view, however, Carstensen (2004) argues that conventional money demand specifications have become unstable, with the break-point from the multivariate break analysis located at the end of 2001, coinciding with the findings of the univariate analysis. This suggests that structural factors more recently may also be relevant and could lead to somewhat different interpretations of monetary developments and their implications for inflation.
51. Key to determining the liquidity (and inflation) implications associated with changing velocity is a better understanding of its root causes—in particular the relative roles of macroeconomic versus structural factors. Disentangling the underlying causes behind “structural” trend breaks in velocity’s time series is crucial—in particular, the extent to which these are long-lasting or fleeting changes, and whether they present risks of higher inflation. The present macroeconomic environment and outlook—i.e., low inflation, interest rates and asset returns—favor holding more money longer, but the component decline in velocity driven by these factors (particularly, expectations) could change relatively swiftly as conditions change. On the other hand, structural factors could indicate more fundamental changes that could endure. Hence, ascertaining the various economic factors (and their relative roles) is central for reliably determining the medium-term inflationary dynamics and risks surrounding current monetary developments.
52. Although firm conclusions are difficult to draw given limited information and the short time elapsed, several tentative explanations may be at work with respect to recently changing liquidity preference in the euro area—beyond traditional money-demand considerations:
Portfolio shifts and heightened risk aversion in the wake of the stock market boom and bust may underlie a protracted flight to safety offered by monetary financial instruments.46
The euro cash changeover created a new monetary reality—including the introduction of large denominations for the first time in several countries—for a large currency union of 300 million people. This may have affected demands for cash, including in the informal sector both within and beyond euro area borders.
The advent of the European single currency has also altered the landscape of international monetary system. The international role of the euro—as vehicle, reserve, and invoice currency—is significant and expanding, suggesting the increasing relevance of currency substitution and foreign demand for euros.47
53. Portfolio shifts seeking the safety of monetary instruments in the wake of financial market turbulence offer a partial explanation for increase demand for liquidity. Analysis conducted by ECB staff posits that increased risk aversion among investors has raised demand for more liquid and more secure financial instruments as found in M3.48 Figure II.9 shows recent growth in M3 and an adjusted series attempting to correct for these portfolio shifts based on ECB estimates.49 As evident in the figure, a portion of the rapid expansion in broad money in the aftermath of the stock market decline in 2000–01 can be attributed to these considerations. Less clear from the analysis is the extent to which this phenomenon should also feature in other industrial countries with similar equity market developments (i.e., common shocks) and whether an unwinding would necessarily lead to higher goods (rather than asset) price inflation. It should also be recognized that even after correcting for the (estimated) impact of portfolio shifts, adjusted broad money has nonetheless persistently grown at a relatively brisk pace of close to 6 percent. This suggests that other explanations also need to be pursued.
M3 and Adjusted M3 Growth, 2000–04
(In percent)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB.M3 and Adjusted M3 Growth, 2000–04
(In percent)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB.M3 and Adjusted M3 Growth, 2000–04
(In percent)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB.54. The euro cash changeover coincides with a rapid increase in currency demand that has yet to subside. In January 2002, the enormous undertaking of introducing physical euro notes and coins ushered in a new monetary era in a very tangible way. Around that time, area-wide currency in circulation displayed an expected “v-shaped” pattern, as legacy currencies (and their demands) were being phased out and replenished by new euro notes and coins; see Figure II.10. Less expected, however, has been the brisk, sustained rise in currency demand (and, hence, M1) ever since the changeover which accounts, to a large extent, for the rapid growth in M3.
Currency in Circulation
(Billions of euros)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB.Currency in Circulation
(Billions of euros)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB.Currency in Circulation
(Billions of euros)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: ECB.55. Initial data suggest cash “hoarding” by agents early in the changeover process, particularly in the largest currency denominations (i.e., €500, €100 notes), a trend which has apparently continued.50 In most member states, the introduction of currency denominations equivalent to the €500 note was unprecedented, perhaps leading to some substitution within demands for M1 instruments but also satisfying some pent-up demands. Rogoff (1998) argues that growing currency holdings (relative to recorded output) in OECD countries reflect vibrant underground economies, where probably greater than 50 percent of banknotes circulate.51 Against this backdrop, the introduction of large euro banknote denominations, in his view, has offered an attractive operating currency for the underground economy, not just domestically but globally. This includes possible growing counterfeiting activities that could have repercussions for money demand for legal tender (i.e., Gresham’s law). In addition, network externalities offered by the advent of a single currency in a monetary union of more than 300 million people is likely to boost its wider usage and demand.52
56. Beyond the informal sector, the international role of the euro continues to expand, with possible implications for money demand. Estimates indicate that one out of every ten euro notes circulates outside the area, still well below estimates (around 50 percent) for the U.S. dollar.53 While the figure is not trivial in the case of the euro, it comprises a small part of the stock of broad money.54 However, measurement problems since the changeover make current estimates more uncertain.55 Meanwhile, net currency shipments of euros abroad continue, and the single currency is playing an expanding role on the international stage.56 As a store of value, euro reserves have been steadily growing as a share of world currency reserves (mainly at the expense of other non-dollar currencies).57 As a vehicle or reference currency in the 2-trillion dollar daily foreign exchange market, 98 percent of all foreign exchange transactions involve either the dollar or the euro, with the euro-dollar exchange rate as the most heavily traded currency pair (approximately $500 billion per day).58
57. As an emerging rival to the U.S. dollar as an international currency, the euro is likely to see increasing currency substitution and foreign demand, influencing area velocity. A glance at velocity’s gap, obtained as the residual from breaking-trend estimates, displays strong correlations with movements in the euro-dollar exchange rate.59 When the euro was relatively weak in value against the U.S. dollar, the velocity of circulation was relatively high (compared to trend); when the euro substantially strengthened and after physical euro notes and coins were introduced, turnover has been relatively low.60
Area Velocity Gap and Dollar-Euro Exchange Rate
(In logarithms unless noted otherwise)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: IFS and staff estimates.1Dollars per euro (LHS).Area Velocity Gap and Dollar-Euro Exchange Rate
(In logarithms unless noted otherwise)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: IFS and staff estimates.1Dollars per euro (LHS).Area Velocity Gap and Dollar-Euro Exchange Rate
(In logarithms unless noted otherwise)
Citation: IMF Staff Country Reports 2005, 266; 10.5089/9781451813029.002.A002
Source: IFS and staff estimates.1Dollars per euro (LHS).G. Concluding Remarks
58. This chapter has sought to examine key issues and features involving monetary aggregates in the euro area as they pertain to the role of money within the ECB’s policy framework. In attempting to ascertain the operational usefulness of money for robustly guiding and communicating policy decisions, the analysis yields the following conclusions:
The nexus between inflation and money growth in the euro area appears most reliable over longer horizons. This delineates a useful role for money as a long-run anchor within the monetary policy strategy, though its shorter-term operational role is less apparent based on the available data and made less clear by the recent dynamics in money’s velocity of circulation.
Area-wide velocity appears to have experienced a variable rate of decline, challenging a robust assessment of liquidity. “Excess” money growth thus far has largely found its counterpart, not in higher prices or output, but in a significant trend decline in velocity. Structural change analysis suggests several structural breaks in trend velocity as late as 2000–01, and recent trend estimates for velocity’s decline lie outside the assumed range that is consistent with the ECB’s reference value for M3 growth. In addition to data uncertainties, this potentially weakens the reliability of static reference values, liquidity measures, and their implications for inflation risks.
Strengthening the cross check as an indicator of underlying inflationary pressures entails further determination and disentangling of the underlying sources of changing liquidity preference. A more definitive understanding of the forces and dynamics behind velocity changes would strengthen the assessment of liquidity conditions and their inflationary consequences. At issue is determining the extent to which macroeconomic versus structural factors account for recent changes, whether these developments could rapidly reverse, and whether this might pass through to higher HICP inflation. The ongoing challenge is to do so, and to accurately draw and convey the attendant monetary policy implications, with limited information as these developments are still unfolding.
Changing liquidity preference could partly reflect fundamental changes stemming from the creation of a large currency union. Following the advent of Europe’s single currency, the physical changeover to euro notes and coins coincided closely with a sharp rise in currency demand that has yet to subside. The euro’s expanding role as an international currency—as an eventual rival to the U.S. dollar—also portends increasing foreign demand. Wide swings in the value of the euro against the dollar, for example, have also correlated strongly with changing velocity, suggesting the importance of currency substitution and non-resident demands.
APPENDIX II A
Unit Root Tests of Euro Area Velocity, 1980:Q1-2004:Q3
Augmented Dickey-Fuller/-test for unit root against trend-stationary alternative; lag length (=1) selected based on Akaike information criterion (AIC).
Kiwatowksi, et al (1992) r|(x) test for trend stationarity against unit root alternative; lags=1.
Unit Root Tests of Euro Area Velocity, 1980:Q1-2004:Q3
| Variable | ADF1 | KPSS2 |
|---|---|---|
| v | -1.89 | 0.23** |
| Δv | -6.60** | 0.11 |
Augmented Dickey-Fuller/-test for unit root against trend-stationary alternative; lag length (=1) selected based on Akaike information criterion (AIC).
Kiwatowksi, et al (1992) r|(x) test for trend stationarity against unit root alternative; lags=1.
Unit Root Tests of Euro Area Velocity, 1980:Q1-2004:Q3
| Variable | ADF1 | KPSS2 |
|---|---|---|
| v | -1.89 | 0.23** |
| Δv | -6.60** | 0.11 |
Augmented Dickey-Fuller/-test for unit root against trend-stationary alternative; lag length (=1) selected based on Akaike information criterion (AIC).
Kiwatowksi, et al (1992) r|(x) test for trend stationarity against unit root alternative; lags=1.
Structural Break Analysis of Euro Area Velocity, 1980:Q1-2004:Q3
(Trend Stationary Model: vt = μ + γt + εt)
Coefficient on time trend; at annual rate, in percent.
Test for parameter stability based on Hansen (1991).
Quandt-Andrews structural change test; critical values based on Andrews (1993).
Structural change test based on Andrews and Ploberger (1994);
Multiple break test based on Bai and Perron (2003).
Unit root tests based on Phillips and Perron (1988) and Perron (1997).
Structural Break Analysis of Euro Area Velocity, 1980:Q1-2004:Q3
(Trend Stationary Model: vt = μ + γt + εt)
| Model: | Linear Trend | Breaking Trend | |||
|---|---|---|---|---|---|
| Slope1 | -1.07** | -2.59** (post break) | |||
| Stability1 | 0.49* | 0.02 (post break) | |||
| R2 | 0.94 | 0.97 | |||
| Structural Break Tests | Test Statistic | Breakdate | Test Statistic | Other Breakdates | |
| Maximal F3 | 110.1** | 2001Q4 | 32.4** | 1994Q2 | |
| Sup.LM4 | 24.5** | 2001Q1 | 22.2** | 1994Q4 | |
| Exp. Avg LM4 | 8.0** | ” | 8.0** | ” | |
| Bai-Perron5 | ... | 2000Q2 | ... | 1992Q3 | |
| Unit Root Test6 | -1.65 | ... | -3.78 | 2001Q1 | |
Coefficient on time trend; at annual rate, in percent.
Test for parameter stability based on Hansen (1991).
Quandt-Andrews structural change test; critical values based on Andrews (1993).
Structural change test based on Andrews and Ploberger (1994);
Multiple break test based on Bai and Perron (2003).
Unit root tests based on Phillips and Perron (1988) and Perron (1997).
Structural Break Analysis of Euro Area Velocity, 1980:Q1-2004:Q3
(Trend Stationary Model: vt = μ + γt + εt)
| Model: | Linear Trend | Breaking Trend | |||
|---|---|---|---|---|---|
| Slope1 | -1.07** | -2.59** (post break) | |||
| Stability1 | 0.49* | 0.02 (post break) | |||
| R2 | 0.94 | 0.97 | |||
| Structural Break Tests | Test Statistic | Breakdate | Test Statistic | Other Breakdates | |
| Maximal F3 | 110.1** | 2001Q4 | 32.4** | 1994Q2 | |
| Sup.LM4 | 24.5** | 2001Q1 | 22.2** | 1994Q4 | |
| Exp. Avg LM4 | 8.0** | ” | 8.0** | ” | |
| Bai-Perron5 | ... | 2000Q2 | ... | 1992Q3 | |
| Unit Root Test6 | -1.65 | ... | -3.78 | 2001Q1 | |
Coefficient on time trend; at annual rate, in percent.
Test for parameter stability based on Hansen (1991).
Quandt-Andrews structural change test; critical values based on Andrews (1993).
Structural change test based on Andrews and Ploberger (1994);
Multiple break test based on Bai and Perron (2003).
Unit root tests based on Phillips and Perron (1988) and Perron (1997).
Structural Break Analysis of Alternative Velocity Series, 1970:Q1-2004:Q2
(Trend Stationary Model: vt = μ + γt + εt)
Coefficient on time trend; at annual rate, in percent.
Test for parameter stability based on Hansen (1991).
Quandt-Andrews structural change test; critical values based on Andrews (1993).
Structural change test based on Andrews and Ploberger (1994);
Multiple break test based on Bai and Perron (2003).
Unit root tests based on Phillips and Perron (1988) and Perron (1997).
Structural Break Analysis of Alternative Velocity Series, 1970:Q1-2004:Q2
(Trend Stationary Model: vt = μ + γt + εt)
| Model: | Linear Trend | Breaking Trend | |||
|---|---|---|---|---|---|
| Slope1 | -0.64** | -2.08** (post break) | |||
| Stability2 | 3.70** | 0.56* (post break) | |||
| R2 | 0.57 | 0.94 | |||
| Structural Break Tests | Test Statistic | Breakdate | Test Statistic | Other Breakdates | |
| Maximal F3 | 901.6** | 1987Q2 | 61.4**, 32.4** | 1972Q4, 2001Q2 | |
| Sup.LM4 | 90.1** | 1992Q2 | 43.1**, 29.4** | 1989Q1, 2001Q3 | |
| Exp. Avg LM4 | 41.7** | ” | 17.9**,12.5** | ” | |
| Bai-Perron5 | ... | 1986Q1 | ... | 1992Q3, 2000Q4 | |
| Unit Root Test6 | -0.33 | ... | -3.75 | 1987Q2 | |
Coefficient on time trend; at annual rate, in percent.
Test for parameter stability based on Hansen (1991).
Quandt-Andrews structural change test; critical values based on Andrews (1993).
Structural change test based on Andrews and Ploberger (1994);
Multiple break test based on Bai and Perron (2003).
Unit root tests based on Phillips and Perron (1988) and Perron (1997).
Structural Break Analysis of Alternative Velocity Series, 1970:Q1-2004:Q2
(Trend Stationary Model: vt = μ + γt + εt)
| Model: | Linear Trend | Breaking Trend | |||
|---|---|---|---|---|---|
| Slope1 | -0.64** | -2.08** (post break) | |||
| Stability2 | 3.70** | 0.56* (post break) | |||
| R2 | 0.57 | 0.94 | |||
| Structural Break Tests | Test Statistic | Breakdate | Test Statistic | Other Breakdates | |
| Maximal F3 | 901.6** | 1987Q2 | 61.4**, 32.4** | 1972Q4, 2001Q2 | |
| Sup.LM4 | 90.1** | 1992Q2 | 43.1**, 29.4** | 1989Q1, 2001Q3 | |
| Exp. Avg LM4 | 41.7** | ” | 17.9**,12.5** | ” | |
| Bai-Perron5 | ... | 1986Q1 | ... | 1992Q3, 2000Q4 | |
| Unit Root Test6 | -0.33 | ... | -3.75 | 1987Q2 | |
Coefficient on time trend; at annual rate, in percent.
Test for parameter stability based on Hansen (1991).
Quandt-Andrews structural change test; critical values based on Andrews (1993).
Structural change test based on Andrews and Ploberger (1994);
Multiple break test based on Bai and Perron (2003).
Unit root tests based on Phillips and Perron (1988) and Perron (1997).
Structural Break Analysis of Alternative Velocity Series, 1970:Q1-2004:Q2
(Difference Stationary Model: Δvt = μ+ εt)
Point estimate on constant term; at annual rate, in percent.
Test for parameter stability based on Hansen (1991).
Based on F-distribution.
Breakdate(s) based on minimized sum of squares a la Bai and Perron (2003).
Non-parametric unit root test based on Phillips and Perron (1988).
Structural Break Analysis of Alternative Velocity Series, 1970:Q1-2004:Q2
(Difference Stationary Model: Δvt = μ+ εt)
| Model: | First Differences | Mean Shift | |||
|---|---|---|---|---|---|
| Drift1 | -0.81** | -2.11** (post break) | |||
| Stability2 | 1.07** | 0.18 (post break) | |||
| R2 | 0.00 | 0.09 | |||
| Structural Break Tests | Test Statistic | Breakdate | Test Statistic | Other Breakdates | |
| Maximal F3 | 13.6** | 1986Q4 | 5.4**, 7.9** | 1972Q4, 2001Q1 | |
| Minimum SSR4 | ... | 1986Q3 | ... | 1993Q2, 2000Q4 | |
| Unit Root Test5 | -9.01** | ... | ... | ... | |
Point estimate on constant term; at annual rate, in percent.
Test for parameter stability based on Hansen (1991).
Based on F-distribution.
Breakdate(s) based on minimized sum of squares a la Bai and Perron (2003).
Non-parametric unit root test based on Phillips and Perron (1988).
Structural Break Analysis of Alternative Velocity Series, 1970:Q1-2004:Q2
(Difference Stationary Model: Δvt = μ+ εt)
| Model: | First Differences | Mean Shift | |||
|---|---|---|---|---|---|
| Drift1 | -0.81** | -2.11** (post break) | |||
| Stability2 | 1.07** | 0.18 (post break) | |||
| R2 | 0.00 | 0.09 | |||
| Structural Break Tests | Test Statistic | Breakdate | Test Statistic | Other Breakdates | |
| Maximal F3 | 13.6** | 1986Q4 | 5.4**, 7.9** | 1972Q4, 2001Q1 | |
| Minimum SSR4 | ... | 1986Q3 | ... | 1993Q2, 2000Q4 | |
| Unit Root Test5 | -9.01** | ... | ... | ... | |
Point estimate on constant term; at annual rate, in percent.
Test for parameter stability based on Hansen (1991).
Based on F-distribution.
Breakdate(s) based on minimized sum of squares a la Bai and Perron (2003).
Non-parametric unit root test based on Phillips and Perron (1988).
References
Andrews, D., (1993), “Tests for parameter instability and structural change with unknown change point,” Econometrica, 61, 821–856.
Bai, J., and P. Perron (2003), “Computation and Analysis of Multiple Structural Change Models,”Journal of Applied Econometrics, pp 1–22.
Bai, J., and P. Perron (1998), “Estimating and Testing Linear Models with Multiple Structural Changes,”Econometrica, Vol. 66 (1), pages 47–78.
Bank of International Settlements, “Triennial Central Bank Survey Foreign Exchange and Derivatives Market Activity in 2004,” March 2005.
Benhabib, J., S. S. Schmitt-Grohe, and M. Uribe (2001), “The Perils of Taylor Rules,” Journal of Economic Theory, 96 (1-2): 40–69.
Brand, C. and N. Cassola (2000), “A Money Demand System for Euro Area M3,”ECB Working Paper Series, No. 39.
Brand, C., D. Gerdesmeier and B. Roffia (2002), “Estimating the Trend of M3 Income Velocity Underlying the Reference Value for Monetary Growth,” ECB Occasional Paper Series, No. 3.
Bruggeman, A. (2000), “The stability of EMU-wide money demand functions and the monetary policy strategy of the European Central Bank,” Manchester School 68, 184–202.
Bruggeman, A., P. Donati and A. Warne (2003), “Is the Demand for Euro Area M3 stable?” in: O. Issing (ed.), Background Studies for the ECB’s Evaluation of its Monetary Policy Strategy, ECB: Frankfurt, pp. 245–300.
Calza, A., D. Gerdesmeier and J. Levy (2001), “Euro area money demand: Measuring the opportunity costs appropriately, IMF Working Paper No. 01 /179.
Calza, A., and J. Sousa (2003), “Why has broad money demand been more stable in the euro area than in other economies? in: O. Issing (ed.), Background Studies for the ECB’s Evaluation of its Monetary Policy Strategy, ECB: Frankfurt, 229–243.
Carstensen, K., (2004), “Is European Money Demand Still Stable?” Working Papers 1179, Kiel Institute for World Economics (revised).
Carstensen, K.,, (2003), “Estimating the ECB policy reaction function,” German Economic Review, forthcoming.
Cassola, N., and C. Morana (2002), “Monetary policy and the stock market in the Euro area,”ECB Working Paper No. 119.
Coenen, G., and J.-L. Vega (2001), “The demand for M3 in the Euro area,” Journal of Applied Econometrics, 16, 727–748.
Drehmann, M., C. Goodhart, M. Kruger, “The Challenges Facing Currency Usage: Will the Traditional Transaction Medium be Able to Resist Competition From the New Technologies?” Economic Policy, Vol. 17, pp. 193–227, 2002
European Central Bank (2005), Review of the International Role of the Euro.
European Central Bank (2004a), The Monetary Policy of the ECB, (2004).
European Central Bank (2004b) “Monetary Analysis in Real Time,” ECB Monthly Bulletin October 2004, pp. 43–66.
European Central Bank (2003) “Estimating the size of portfolio shifts from equity to money” Box 1 of the ECB Monthly Bulletin May 2003, pp. 11–14.
Fagan, G., Henry, J. and R. Mestre (2001), “Area-Wide Model (AWM) for the Euro Area,” ECB Working Paper Series,. 42.
Feige, E. (1996), ”Overseas Holdings of U.S. Currency and the Underground Economy,” in S. Pozo (ed.), Exploring the Underground Economy: Studies of Illegal and Unreported Activity, Kalamazoo, Michigan: W.E. Upjohn Institute for Employment Research, pp. 5–62.
Feige, E. and J. Dean (2004), “Dollarization and Euroization in Transition Countries: Currency Substitution, Asset Substitution, Network Externalities and Irreversibility” in V. Alexander, J. Mélitz, and G. von Furstenberg (eds.), Monetary Unions and Hard Pegs: Effects on Trade, Financial Development, and Stability, Oxford University Press, 2004:p. 438–464
Fischer, B., P. Köhler, and F. Seitz (2004), “The Demand for Euro Area Currencies: Past, Present and Future,” ECB Working Paper No 330.
Galí, J., S. Gerlach, J. Rotemberg, H. Uhlig and M. Woodford (2004), The Monetary Policy Strategy of the ECB Reconsidered: Monitoring the European Central Bank 5, CEPR: London.
Gerlach, S. and L. Svensson (2003), “Money and Inflation in the Euro Area: A Case for Monetary Indicators?” Journal of Monetary Economics, 50, 1649-1672.
Gerlach, S. (2004), ”Interest Rate Setting by the ECB: Words and Deeds,” CEPR Discussion Paper No. 4775.
Gerlach, S., (2004), “The Two Pillars of the ECB.” Economic Policy, 2004, 40, 389–439.
Gros, D., T. Mayer, and A. Ubide (2005), EMU at Risk, 7th Annual Report of the CEPS Macroeconomic Policy Group.
Hansen, B., (2001), “The New Econometrics of Structural Change,” Journal of Economic Perspectives.
Hansen, B., (1991), “Parameter Instability in Linear Models,”Journal of Policy Modeling 14(4), pp. 517–533.
International Monetary Fund(2004), Annual Report of the Executive Board for the Financial Year ending April 30, 2004, IMF: Washington, D.C.
Jäger, A., (2003), “The ECB’s money pillar: An assessment,”IMF Working Paper, No. 03/82.
King, M., (2001), “No Money, No Inflation—The Role of Money in the Economy,” Bank of England Quarterly Bulletin, pp.162–77.
Kontolemis, Z., (2002) ‘Money Demand in the Euro Area: Where Do We Stand. Today?’. IMF Working Paper No. 02/185.
Masuch, K., S. Nicoletti-Altimari, H. Pill and M. Rostagno (2003) The Role of Money in Monetary Policy Making, in O. Issing (ed.), Background Studies for the ECB’s Evaluation of its Monetary Policy Strategy, ECB: Frankfurt, pp. 187–228.
McCallum, B., (2002), “Monetary Policy Analysis,” NBER Reporter (Spring).
Nicoletti-Altimari, S. (2001), “Does money lead inflation in the euro area?” ECB Working Paper No. 63.
Orphanides, A., (2003a), “The Quest for Prosperity without Inflation,” Journal of Monetary Economics. 50(3): 633–663.
Orphanides, A., (2003b), “Monetary Policy Evaluation with Noisy Information,” Journal of Monetary Economics, Volume 50, Issue 3, pp. 605–631.
Perron, P., (1997), “Further evidence on breaking trend functions in macroeconomic variables,” Journal of Econometrics, 80, 355–385.
Phillips, P., and P. Perron (1988), “Testing for a Unit Root in Time Series Regressions,” Biometrika.
Porter, Richard, and Judson, Ruth. (1996) “The Location of US Currency: How Much is Abroad?” Federal Reserve Bulletin, October 82, pp. 883–903.
Rogoff, K. (1998), “Blessing or Curse? Foreign and Underground Demand for Euro Notes,” Economic Policy, 13, pp. 263–303.
Rogoff, K. (2002), “The Surprising Popularity of Paper Currency,” Finance and Development, March, Volume 39, Number 1.
Schneider, F., and D. Enste, (2005), The Shadow Economy: An International Survey Cambridge Press.
Prepared by Hamid Faruqee (EUR).
Price stability for the ECB is defined in the context of year-on-year increases in the Harmonized Index of Consumer Prices (HICP); a “medium term” orientation is meant to convey a forward-looking dimension to monetary policy decisions, although no explicit timeframe is cited given the uncertainties involved; see ECB (2004).
See, for example, Gerlach (2004), Gali et al (2004).
ECB’s October 2004 press statement.
ECB (mid-point) projections for growth in 2005 were marked down from 2.3 percent in September 2004 to 1.6 percent in March 2005; annual HICP inflation (mid-point) projections for 2005 held steady, but was expected to further decline to 1.6 percent in 2006.
See Chapter III for a discussion of asset prices and their operational role for monetary policy.
See ECB Monthly Bulletin(October 2004).
The reference value for money does not entail a commitment on the part of the ECB to mechanically target monetary growth. Rather, monetary developments in light of the reference value are analyzed, in conjunction with other indicators, in order to ascertain their implications for inflation. In May 2003, the Governing Council decided to no longer annually review its underlying medium-term reference value, but would continue to monitor the validity of these underlying assumptions as necessary.
See ECB (2004). The ECB opted not to announce money’s reference value as a range to avoid misconceptions that this would constitute an operational target for monetary policy.
Data on velocity shown in the figure are based on the ECB’s “preferred” series derived from aggregated series for money, output and prices based on irrevocable exchange rates for the period prior to EMU. Data issues and alternative series are discussed later in Section E.
Analyzing data from 1980-2001, Gerlach and Svensson (2003) find that the real money gap is useful for forecasting inflation, while the standard money growth indicator has no predictive power, despite stable money demand. For data between 1970 and 2003, Gerlach (2004) finds that money growth is informative for future inflation, on par with the usefulness of the output gap. In the pre-EMU era, Jaeger (2003) finds that both the money and output gaps predict inflation—the former especially over longer horizons.
Equivalently, the income elasticity of money demand is typically found to be greater than unity; see Masuch et al. (2003) and Brand et al. (2002) for a review. But note that this is simply the other side of the same coin, since (residual) velocity is imposed to be stationary in these estimates and, in any case, cannot be observed independently.
See Fischer and others (2004) for a discussion of national differences regarding money.
See, for example, Bruggeman and others (2003). While recognizing that no method is clearly superior, they favor this approach as it (i) creates a “synthetic” euro area historically using the same technique applied after the start of EMU, and (ii) preserves balance sheet identities. However, this approach cannot be extended to interest rates, thus requiring a “mixed” aggregation method in order to study area-wide money demand.
Specifically, national variables, including (log) GDP, are aggregated using the “index method” with fixed weights for countries based on their share in constant GDP measured at PPP exchange rates in 2001; see Fagan et al (2001). This approach is predicated (subject to the usual caveats) on the assumption that PPP provides useful platform to make international comparisons—e.g., living standards and real production across countries.
Building on this premise and targeting U.S. benchmarks for money and credit, Gros and others (2005) warn that instead of converging toward U.S. levels, money and bank credit growth in the euro area have, in fact, matched or exceeded that in the United States. A short time perspective (i.e., since EMU), however, belies several underlying financial trends.
Moreover, data on broad money and nominal GDP for the United States, United Kingdom, and Canada confirm that velocity’s secular decline is a more widespread phenomenon.
This applies to narrow money as well, including currency in circulation. After the arrival of automated teller machines (ATMs), other innovations include debit cards, electronic payments, smart cards, and network or e-money. Nevertheless, the adage “cash is king” still holds true, not least of all due to the anonymity it provides; see Drehman et al (2002) and Rogoff (2002).
Stability tests based on Hansen (1991). Parameter constancy is rejected at the 1 percent level in the longer velocity series; see appendix.
The critical values are based on Andrews (1993). To avoid end-point sensitivity, the series were trimmed (by either 10 or 15 percent) on both ends in applying various break tests; note that the maximal F-test statistic accurately dates the timing of the break only under certain regularity conditions; see Hansen (2001).
Selection of the breakdate is chosen using Perron’s “additive outlier model”; see Perron (1997); see Table A2.
The sequential method—i.e., further splitting the sample based on the break(s) found at the previous step—quickly loses degrees of freedom and may lead to detection of spurious breaks. The Bai and Perron (1998, 2003) approach considers the number and location of breaks simultaneously based on the entire sample to address spurious breaks. In terms of the number of breaks, the Bayesian Information Critierion (BIC), penalizing additional parameters (i.e., breaks)—favors the one-break model over the two-break model in the shorter series.
Based on data from 1970-2003, Gerlach (2004) finds that money growth’s ability to help predict future inflation changes (i.e., weakens) from the pre-1987 period to the later period.
Adding an autoregressive coefficient on lagged velocity growth would yield a coefficient around 0.3; the estimated long-run drift in the series and parameter instability issues are unaffected.
The growth model for velocity is given by
Using HICP inflation (instead of the GDP deflator) to construct a long series for velocity would yield a corresponding reference value around 5–5¼ percent.
See, for example, Coenen and Vega (1999), Brand and Cassola (2000), Calza and others (2001). More recent studies broadly supporting stable area-wide money demand (till 2001) are Brand et al (2002) and Bruggeman et al (2003). Kontolemis (2002) shows, though, that periods of disequilibrium arising from velocity shocks, can be protracted.
Akin to ECB staff analysis, Carstensen (2004) emphasizes stock market variables and portfolio adjustment to explain the money demand instabilities identified in his analysis.
See ECB (2005) for a review of the euro’s international role. Note that M3 is a resident-based concept—i.e., money held by agents residing in the euro area. However, currency substitution (by residents) and measurement difficulties in tracking euros held by non-residents can allow international influences to play a role. See discussion below.
See ECB, Monthly Bulletin, May 2003 and October 2004. The ECB’s monetary analysis also examines the role of institutions and financial innovation in assessing liquidity conditions in real time. For example, the growth of internet banks—offering attractive deposit rates—has garnered recent attention by ECB staff.
The corrected series, derived from univariate estimates augmented by crude intervention variables (e.g., dummies), exhibit portfolio shifts as early as 1992, but a zero impact on many intervening dates. The estimates, as acknowledged in the ECB Monthly Bulletin of October 2004, should be interpreted with caution. Structural change tests still find possible breaks in the corresponding velocity series.
“In contrast to the decline in the number of small banknotes in circulation, the demand for large denominations increased steadily in the first eight months of 2002. The strongest rise in demand was for the EURO 500 banknotes, which more than doubled between mid-January and the end of August [+120%]. Demand also rose during this period for the EURO 100 banknotes, by no less than 60%. This shows that there was a rapid increase in the hoarding of banknotes in the first months of the euro cash era...”—W. Duisenberg (ECB President), speech on the occasion of the International Imaging Industry Summit, (September 2002).
A large literature ubiquitously uses currency demand as the key measure of the relative size of the underground economy; See, for example, Schneider and Enste (2005) and the references cited therein.
See, for example, Feige and Dean (2004).
See, for example, Porter and Judson (1996) and Feige (1996)).
Fischer et al (2004) estimates foreign circulation of euro legacy currencies between 8 and 15 percent in the late 1990s, in line with current estimates of euros circulating abroad.
“As the initial supply of euro banknotes abroad, the so-called frontloading, was entirely channeled in December 2001 through central banks and commercial banks, the initial amount of banknotes that foreigners received was meticulously captured in data... However, as euro banknotes have been available to anyone inside the euro area since then, they can flow out through many unrecorded channels. As time goes by... data on banknotes shipments by banks will become increasingly less reliable as a measure of the foreign circulation of the euro.”—T. Padoa-Schioppa (ECB Board Member), lecture at ECB 8th Dubrovnik Economic Conference (June 2002).
See ECB (2005).
See IMF Annual Report (2004).
See BIS (2005).
The correlation coefficient is -0.86. Velocity gaps shown in the figure are derived from the longer series based on a single trend break (in 1987). The resulting gaps based on this historical break thus encompass both transitory deviations and more recent trend breaks—the latter being particularly relevant for comparisons with candidate factors generating possible structural change in money demand.
Examining demands for euro legacy currencies (and excluding the period directly surrounding the cash changeover), Fischer and others (2004) find a stable long-run relationship between real currency, real private consumption, an opportunity cost variable and the euro’s real effective exchange rate—capturing non-resident demands.
