Rebalancing and the Euro

Abstract

Pushing down the euro is seen by many as a solution to stagnation and ‘lowflation’ in the euro area, but this paper argues it needs to be supported by a fiscal expansion, in particular in Germany. Unfortunately the incentives for Germany to embark on fiscal stimulus are weak. Yet without a fiscal expansion the current account surplus would widen further in response to exchange rate depreciation, thus exacerbating the current imbalances—excess saving in Germany alongside indispensable deleveraging in southern Europe. It would risk producing a protracted slump that feeds onto itself, with (core) inflation declining further and the risk of outright deflation increasing. If that risk were to materialise the ECB may have to be called to rescue the euro again, as it did in 2012 with the introduction of Outright Monetary Transactions (OMT). And it would have to do so persuasively or else Europe risks ending up in a Japanese conundrum. It is in this perspective the ECB's move towards quantitative easing should be seen.

The views expressed in this chapter are those of the author only and cannot be attributed to Autonomy Capital or its staff.

Appendices

Appendix: Econometrics

1.1 Methodology and Data

The aim of model is to help identify empirically for the euro area the potential links between the real effective exchange rate (RE), the current account position (CA) and the relative stances of macroeconomic policy, i.e. fiscal policy as gauged by the budget balance as a share of GDP relative to that of trading partners (FB) and monetary policy as gauged by the real interest rate differential (IR). The basic premise is that RE, CA, FB and IR are mutually dependent variables, with the direction of their dependencies in line with basic macroeconomic theory.

If we accept this premise, we can test whether—within the context of a linear vector error-correction model (VECM)—four co-integration vectors exist for the variables of interest given by:

$$RE_{t} = \sum\limits_{i = 1}^{n} {\alpha_{RE,i} X_{i,t} } + \mu_{RE,t}$$

(1)

$$CA_{t} = \sum\limits_{i = 1}^{n} {\alpha_{CA,i} X_{i,t} } + \mu_{CA,t}$$

(2)

$$FB_{t} = \sum\limits_{i = 1}^{n} {\alpha_{FP,i} X_{i,t} } + \mu_{FB,t}$$

(3)

$$IR_{t} = \sum\limits_{i = 1}^{n} {\alpha_{IR,i} X_{i,t} } + \mu_{IR,t}$$

(4)

with the following error-correction equations:

$$\Delta RE_{t} = \theta_{RE} \mu_{RE,t - 1} + \pi_{RE} \mu_{CA,t - 1} + \vartheta_{RE} \mu_{FB,t - 1} + \rho_{RE} \mu_{IR,t - 1} + \sum\limits_{i = 1}^{n + 4} {\sigma_{RE,i}\Delta Y_{i,t - 1} } + \varepsilon_{RE,t}$$

(5)

$$\Delta CA_{t} = \theta_{CA} \mu_{RE,t - 1} + \pi_{CA} \mu_{CA,t - 1} + \vartheta_{CA} \mu_{FB,t - 1} + \rho_{CA} \mu_{IR,t - 1} + \sum\limits_{i = 1}^{n + 4} {\sigma_{CA,i}\Delta Y_{i,t - 1} } + \varepsilon_{CA,t}$$

(6)

$$\Delta FB_{t} = \theta_{FB} \mu_{RE,t - 1} + \pi_{FB} \mu_{CA,t - 1} + \vartheta_{FB} \mu_{FP,t - 1} + \rho_{FB} \mu_{IR,t - 1} + \sum\limits_{i = 1}^{n + 4} {\sigma_{FB,i}\Delta Y_{i,t - 1} } + \varepsilon_{FB,t}$$

(7)

$$\Delta IR_{t} = \theta_{IR} \mu_{RE,t - 1} + \pi_{IR} \mu_{CA,t - 1} + \vartheta_{IR} \mu_{FB,t - 1} + \rho_{IR} \mu_{IR,t - 1} + \sum\limits_{i = 1}^{n + 4} {\sigma_{IR,i}\Delta Y_{i,t - 1} } + \varepsilon_{IR,t}$$

(8)

in which:

$$\Delta X_{j,t} = \sum\limits_{i = 1}^{n + 4} {\sigma_{X,i,j}\Delta Y_{i,t - 1} } + \varepsilon_{X,j,t}$$

(9)

where Y denotes all endogenous variables in the system, comprising the variables of interest RE, CA, FB, IR and a set of other variables X. Furthermore, μ denotes the error correction terms and ε the residuals in the error-correction equations. Note that in this system fiscal and monetary policy is endogenous. In the case of fiscal policy this could be attributed to the automatic fiscal stabilisers or, more broadly, rules-based fiscal policy responses. In the case of monetary policy this could be interpreted as policy reactions following some unspecified policy reaction function. Purely ‘discretionary’ fiscal and monetary policy moves are then captured by the ‘innovations’ $\varepsilon_{FB,t}$ and $\varepsilon_{IR,t}$. The parameters of prime interest are those on the error-correction terms in the error-correction equations, where we expect:

$$\theta_{RE} < 0,\;\theta_{CA} < 0, \pi_{RE} > 0, \pi_{CA} < 0,\;\rho_{RE} > 0, \vartheta_{FB} < 0, \vartheta_{CA} > 0 \;and\;\rho_{IR} < 0.$$

(10)

This effectively says that all four co-integrated variables will (by definition) converge to their long-run equilibrium, that an excessive current account surplus will over time produce an appreciation of the real exchange rate, that an overvalued real exchange rate will over time cause the current account surplus to shrink, that a fiscal policy tightening will lead the current account surplus to increase and that a tight stance of monetary policy will cause the real exchange rate to appreciate. Other relationships in the system may be more ambiguous.

The selected variables X are standard determinants of the equilibrium real effective exchange rate, comprising (for sources and definitions see Appendix):

• TNT = Ratio of prices of tradable to non-tradable goods (relative to trading partners)

• TOT = Terms of trade (relative to trading partners)

• PROD = Real GDP per capita (relative to trading partners)

• NFA = ECB net foreign assets as a share of GDP

This choice of explanatory variables will allow us to interpret RE as predicted by Eq. (1) as the equilibrium real effective exchange rate. The variables TNT and PROD are gauges of the Balassa-Samuelson effect, TOT aims to capture the impact of commodity price shocks and NFA captures the impact of currency intervention on RE. All variables are natural logarithms except for CA, FB, NFA, which are shares in GDP, and IR which is defined in percentage-points/100. The definitions and sources are:

RE = Real Effective Exchange Rate using the CPI of 37 trading partners as the deflator (Eurostat)

IR = Real short term interest rate differential (euro area—weighted average OECD, OECD)

FB = Net lending of General Government as a % of GDP, differential against OECD average, OECD)

TNT = Measured as the (EA PPI)/(EA HICP), EA PPI = Total output price index of industry excluding construction, sewerage, waste management, and remediation activities (Eurostat), EA HICP = All-items HICP (Eurostat)

TOT = Export Price Index/Import Price Index, Export Price Index of goods and services (Eurostat), Import Price Index of goods and services (Eurostat)

CA = Quarterly Current Account Surplus/Quarterly GDP (Eurostat)

$${\text{PROD}} = \frac{{\frac{Euro\;Area\;Real\;GDP}{Capita}}}{{\frac{OECD\;Real\;GDP}{Capita}}} \left( {\text{OECD}} \right)$$

NFA = Euro Area Central Bank Net Foreign Assets/Quarterly Euro Area GDP, Euro Area Central Bank NFA from IMF Central Bank Survey

All data is quarterly and available for the period 1999Q1 to 2013Q3.

Estimation Results

The estimation results, reported in Table 1 (based on the data reported in Table 2), are overall satisfactory in the sense that the trace test suggests the existence of four co-integration equations and that all signs are expected as in (10). Specifically,

• all coefficients on the ‘self’ error –correction terms in the short-term equations have the required negative sign and are significant;

• the impact of the error correction term in the current account long-run equation on the change in the real exchange rate is positive and significant;

• the impact of the error correction term in the real exchange rate long-run equation on the change in the current account is negative and significant;

• the impact of the error correction term in the fiscal balance long-run equation on the change in the current account is positive and significant; and

• the impact of the error correction term in the real interest rate differential long-run equation on the change in the real exchange rate is positive, though not significant.

Table 1 Vector error correction estimates

Table 2 Data

Other interesting features of the estimation result regarding the long-run equations are that relative per capita income is found to exert a positive impact on the real effective exchange rate (as expected) and a negative impact on the current account balance. So everything else equal, high income economies should run a smaller current account balance and a higher real exchange rate than low income countries. Moreover, again everything else equal, high income countries run more positive fiscal positions and higher real interest rates than lower income countries per capita income.