您当前的位置:首页 > 研究文萃
殷琳博士撰写的《货币联盟内的价格调整:思考最佳货币区标准内生性问题的新角度》(第二作者)在英文权威期刊《牛津经济统计公报》2018年第4期发表
信息来源:   作者:   发表时间:2018-07-05  阅读次数:134

货币联盟内的价格调整:思考最佳货币区标准

内生性问题的新角度

Michael Bleaney,诺丁汉大学经济学院教授

殷琳,铁道党校科研所

 

摘要:在理性预期模型中,对于经济冲击,货币联盟制度由于不存在汇率调整,相对于可调整的盯住汇率制度应在工资和物价上有更大的反映,即外部冲击对于这类国家有更显著的经济影响。这是最佳货币区标准中经常被忽略的内生性问题。本文通过来自三个货币联盟的经验证据证明了这一观点,数据显示最佳货币区内存在一定的价格调整空间,但调整速度是缓慢的。

    关键词:货币联盟;汇率制度;价格调整;最佳货币区标准

Price Adjustment in Currency Unions:

Another Aspect of the Endogeneity of the

Optimum Currency Area Criteria?

Michael Bleaney1 and Lin Yin2

1School of Economics, University of Nottingham, University Park,  Nottingham NG7 2RD,  UK (e-mail: michael.bleaney@nottingham.ac.uk)

2Economic Research Department, China Railway Party School, Beijing, China (e-mail: 951298673@qq.com)

 

Abstract

In a rational expectations model, wages and prices should respond more to shocks in currency unions than in adjustable pegs because of the absence of exchange rate adjustment.  This is an aspect of the endogeneity of the optimum currency area criteria that has been largely ignored. Empirical evidence from three currency unions tends to suggest some degree of endogeneity of price flexibility, but the rate of adjustment is slow.  Self-selection into currency unions by countries with naturally greater price flexibility does not appear to be a significant factor.

 

Keywords:     currency union, exchange rate, price, optimum currency area.

JEL No.:   F31

Word Count: ****

1Corresponding author: Professor M F Bleaney, School of Economics, University of Nottingham, Nottingham NG7 2RD, UK. e-mail: michael.bleaney@nottingham.ac.uk. Lin Yin: tclinny@163.com.  Tel. +44 115 951 5464.  Fax +44 115 951 4159. The authors wish to thank Mark Roberts for helpful comments on an earlier version of this paper.

1.      INTRODUCTION

The first sentence of Mundell’s (1961) seminal paper on optimum currency areas reads as follows.  “It is patently obvious that periodic balance of payments crises will remain an integral feature of the international economic system as long as fixed exchange rates and rigid wage and price levels prevent the terms of trade from fulfilling a natural role in the adjustment process.”  The subsequent literature has emphasised the importance of international linkages between regions through the extent of trade, symmetry of shocks, labour mobility and fiscal transfer mechanisms as criteria for a currency union. Frankel and Rose (1997) added a new twist to the debate by noting the endogeneity of some of these criteria, showing that the formation of a currency union itself tended to increase intra-union trade and the symmetry of shocks.  Yet Mundell’s paper, as the quote above makes clear, is not about fixed exchange rates as such, but about fixed exchange rates combined with rigid wages and prices.  This second element has figured relatively little in subsequent discussion, no doubt because the existence of these nominal rigidities is taken for granted.

An important component of modern economic theory, the rational expectations revolution, suggests that nominal rigidities are not entirely determined by the institutional structure of the labour market; rather, they are endogenous to the exchange rate regime in place.  If currency unions were to induce significantly greater wage and price flexibility through the exchange rate commitment, that would reduce the employment effects of region-specific shocks and absorb some of the pressure that would otherwise fall on other adjustment mechanisms.  In other words, the point about the endogeneity of the optimum currency area criteria may also apply to wage and price rigidity.  There has been very little empirical work on this issue in general, although it has been an important element of the discussion and analysis of the crisis in the euro area.

Recent history does not support the view that currency unions are immune from shocks that require internal adjustment.  In the euro area, for example, large capital flows from the centre to the periphery collapsed after the global financial crisis, leaving internal real exchange rates severely misaligned ((Jaumotte and Sodsriwiboon, 2010; Shambaugh, 2012).  If nominal rigidities obstruct the required internal price adjustment, the resulting output losses may precipitate a flight from a country’s assets, particularly if the banking system is also exposed to significant losses, as in the euro area.[1]  In second-generation models of currency crises (e.g. Obstfeld, 1996), a crisis can be precipitated if agents perceive that there is a limit to the output cost that the authorities are willing to bear in defence of the currency.  Exactly the same can happen in a currency union, because a severe crisis is likely to put the possibility of leaving the union and devaluing back on the political agenda.

The general perception of the Eurozone crisis is that adjustment in the periphery has been far too slow to prevent large output losses (Eichengreen et al., 2014; Gibson et al., 2014; Honkapohja, 2014).  Ireland may be the poster boy of post-crisis adjustment, having achieved a substantial reduction in relative unit labour costs, but it is regarded as very much the exception that proves the rule (OECD, 2016; Whelan, 2014).

This paper addresses the issue of adjustment within a currency union.  If an individual member country’s wages and prices get too high relative to those of the other members, the situation can only be corrected by adjustment of relative wages and prices.  This is the question that is addressed here: is there more internal wage and price adjustment in currency unions than under other exchange rate regimes, because of the disciplinary effect of the fixed exchange rate?  Perhaps surprisingly, there has been virtually no formal investigation of this issue.

In a rational expectations model, as mentioned above and shown below, wage- and price-setters should be more willing to adjust wages and prices under hard pegs, because the nominal exchange rate is known to be fixed.  The results of our empirical analysis show that, under floating or soft pegs, adjustment works more or less exclusively through the nominal exchange rate, and price adjustment is small.  In currency unions, as the model predicts, price adjustment is statistically significant, but still slow: a ten percent difference in the level of the real effective exchange rate is estimated to be associated with a relative price movement of less than one percent per annum.

The structure of the paper is as follows.  A theoretical model is presented in Section Two.  The empirical model is laid out in Section Three, and data issues and empirical results are discussed in Section Four.  Section Five concludes.

 

2.      AN ILLUSTRATIVE MODEL

The aim of this section is to present a simple model that captures the idea that under rational expectations fixing the exchange rate alters the response of prices to a negative shock.  In the model presented below, the economy is populated by a set of n monopolistically competitive individual producers of a single differentiated good, as in the sticky-price model of Obstfeld and Rogoff (1996, Ch. 10).  Producers like higher output and higher prices, but demand is a declining function of the real exchange rate. The government likes higher output and also a higher nominal exchange rate (to keep inflation down).  The demand curve is subject to a stochastic shock that is observed by all agents.  With a fixed exchange rate, a negative demand shock induces producers to reduce prices, but the government cannot adjust the exchange rate to further protect output.  With an adjustable peg, under a negative shock the government sets the exchange rate lower than they otherwise would, and after producers have set prices, but producers anticipate this and consequently set prices higher than they would in the currency union case.

The model is in the tradition of Barro and Gordon (1983), in the sense that the government’s preferences incorporate an ambitious output target that generates an inflation bias.  The deviation of (the log of) output (y) from its “natural” level () is a decreasing function of (the log of) the real exchange rate (p+e), adjusted for a zero-mean stochastic demand shock (z), where p is the log of the price level, e is the log of the nominal exchange rate (foreign currency units per unit of domestic currency, so that an increase represents an appreciation of the domestic currency) and foreign prices are fixed at one (so their logarithm is zero):

                    q  > 0                                        (1)

We shall assume that producers set prices so as to minimise the following loss function:

                                                        (2)

In equation (2), producers have a target level of output of , which is assumed to be greater than , and a target price of .  The government cares about output and price stability, and has the same target level of output as the producers.  In order to maintain the one-period nature of the model, the desire for price stability is represented as the government’s preference for a higher nominal exchange rate, so that imported inflation is lower.  The government’s loss function is:

                                                         (3)

where  represents the government’s exchange rate target.

The order of events in each period is as follows.  First the demand shock, z, is determined and observed by all agents.  Then the producers set prices, taking the expected exchange rate into account. Finally the government sets the exchange rate, if it is free to do so.  Writing  and substituting from (1) into (2) yields:

                                           (4)

The producers choose p to minimise (4) subject to their expectations of e and their observation of z.

       If E is the expectations operator, the first-order condition is:

                                     (5)

which yields the solution:

                                                   (6)

Prices are set higher, the more ambitious is the price target () relative to the output target (k), and the lower is the elasticity of demand with respect to the real exchange rate ().  Prices are also higher if the expected exchange rate is lower.

A currency union

Consider first the case of a currency union.  Since the exchange rate is fixed, the reaction of prices to the demand shock may be obtained by partially differentiating (6) with respect to z:

                                                                            (7)

 

An adjustable peg

In the case of an adjustable peg, e can be adjusted after prices have been set.  Since the producers have full information about the government’s loss function, however, under rational expectations they can calculate the exchange rate that will be chosen. The first-order condition for the government’s problem is:

                                          (8)

which yields the solution

                                                   (9)

Adding together equations (6) and (9) yields the following solution for the real exchange rate:

                         (10)

So that the real exchange rate response to the shock is:

                                                                  (12)

which is greater than in the case of a currency union as given by (7).  Solving (6) and (9) for p and e reveals that the real exchange rate response to the shock is split equally between prices and the exchange rate:

                                                           (13)

Equation (13) implies that prices adjust less under an adjustable peg than in a currency union.

 

Summary

In a rational expectations model prices exhibit more flexibility in response to shocks than under an adjustable peg, but the real exchange rate adjusts less and output is more affected by the shock, because the nominal exchange rate cannot perform the role of a shock absorber.

 

 

3.      THE EMPIRICAL MODEL

The model’s three predictions are: comparing hard pegs with soft pegs, (1) inflation is lower; (2) output is more sensitive to shocks; and (3) adjustment to negative shocks has to take the form of money wage reductions, whereas in soft pegs real wages can be reduced by unexpected exchange rate depreciation.  We concentrate on the third prediction, since the first is a well-known result (e.g. Bleaney and Francisco, 2005), and the second has been addressed elsewhere (e.g. Ghosh et al., 2015).  The model assumes purchasing power parity, which is known not to hold in the real world; moreover data on money wages are not available for some of the countries in the sample. Consequently in the empirical analysis we concentrate on adjustment in consumer prices (or alternatively GDP deflators).  The issue is whether domestic prices adjust more to shocks when the exchange rate cannot adjust.

       How should shocks be represented empirically? Since we want them to reflect an exchange rate disequilibrium, we use the lagged real effective exchange rate as a measure of this disequilibrium.

We estimate the following equation:

                                   (17)

where  is the consumer price index in country i in year t,  is the same thing for country i’s anchor currency, R is the real effective exchange rate index (an increase representing an appreciation), RA is the real effective exchange index of the anchor currency, D is the first-difference operator and u is a random error.  Equation (17) expresses inflation relative to that in the anchor currency as a function of the lagged level of the real exchange rate of country i. and the current real exchange rate appreciation of the anchor currency. If b> 0, then there is price adjustment relative to the anchor currency in response to the lagged level of the country’s real effective exchange rate.  In an alternative formulation we replace the consumer price index by the GDP deflator.

 

4.      DATA

There are three currency unions in our analysis: the euro area, the African Financial Community (CFA) and the East Caribbean Currency Authority (ECCA).  The identification of other pegs and floats is drawn from the classification scheme of Bleaney and Tian (2016).[2] The anchor currency is the US dollar for the ECCA, and the euro in the other two cases (i.e. the real effective exchange rate is that for the euro area as a whole).  The anchor currency prices are respectively US prices and those for the whole euro area.  The anchor is “external” for the CFA but “internal” for the euro area because the CFA franc is pegged to the euro, whereas the euro is a floating currency.

Annual data for 186 countries over the period 1980 to 2014 are used (1999 to 2014 for the euro area and the CFA zone), and are mostly drawn from IMF International Financial Statistics (IFS).[3] The GDP deflators are from the IMF World Economic Outlook (WEO) database, and price indices for the euro area from OECD Economic Outlook.

 

5.      EMPIRICAL RESULTS

It is useful to disaggregate the volatility of the real effective exchange rate of a member of a currency union into separate elements.  Let R be the log of the real effective exchange rate of the country, N the log of its nominal effective exchange rate, and P and PF respectively the log of domestic prices and of the trade-weighted average of foreign prices (PA and PFA respectively for the anchor currency).  By the definition of R,

R = N + P – PF                                                                (18)

 and for the anchor currency (A),

RA= NA + PA – PFA                                                            (19)

Subtracting (18) from (17) and taking first differences yields:

DR = DRA + (DN – DNA) + (DP – DPA) – (DPF – DPFA)                               (20)

So real exchange rate movements (DR) can be decomposed into (1) real effective exchange rate changes of the anchor currency (DRA); (2) nominal effective exchange rate changes relative to the anchor currency (DN – DNA), which can only happen (in the absence of a currency union devaluation) because of differences in trade weights; (3) inflation relative to the anchor currency (DP – DPA); and (4) differences in trade-weighted foreign inflation, which again is a matter of weights.  The second and last terms are the effect of different trading partner weights between the member of the currency union and the anchor currency.

       Table 1 shows that real exchange rate volatility of the anchor currency (DRA) has tended to be more important than inflation differentials (DP – DPA) to the real exchange rate volatility of currency union members.  This suggests that the exogenous shocks assumed in the model have been a significant component of real exchange rate volatility in currency unions.


 

 

Table 1. Real Effective Exchange Rate Volatility in Currency Unions

 

Standard deviation

ECCA

1980-2014

CFA

1999-2014

Euro Area

1999-2014

DR

0.0414

0.0461

0.0268

DRA

0.0571

0.0579

0.0540

DN – DNA

0.0495

0.0273

0.0312

DP – DPA

0.0182

0.0297

0.0108

DPF – DPFA

0.0383

0.0104

0.0055

Observations

204

192

226

Anchor currency

US dollar

euro

euro

Notes.  Data are annual. DR (DRA): change in log of real effective exchange rate of member countries (anchor currency).  DN (DNA): change in log of nominal effective exchange rate of member countries (anchor currency).  DP (DPA): change in log of consumer price index of member countries (anchor currency).  DPF (DPFA): trade-weighted average of change in log of trading partners’ consumer price indices of member countries (anchor currency). 

      

In the remainder of this paper we investigate whether the price level of member countries responds to the level of the lagged real effective exchange rate.  The model in Section Two predicts that the response will be stronger in currency unions than in soft pegs.

Table 2 shows the results of estimating equation (17) separately for currency unions, other pegs and floats, and then for the three currency unions individually, using the price index and real effective exchange rate (REER) index of the euro area for the CFA and for individual euro-area countries, and of the US dollar for the ECCA.  Where fixed country effects are not significant (as is the case for currency unions), the equation is estimated by OLS (the p-value of this test is given at the foot of Table 2).  For currency unions the point estimate of the coefficient of the lagged REER is -0.0658, with a t-statistic of -7.42, whereas for other pegs the coefficient is only -0.0103, with a t-statistic of -1.62.  Thus in currency unions there is a significant negative correlation between the lagged REER and consumer price inflation relative to the anchor currency.  This is consistent with the predictions of the model, but it also implies that adjustment to real exchange rate shocks is slow, with a relative inflation effect of a 10 percent difference in the real exchange rate of less than 0.7 percent p.a.

In the individual currency unions the picture is fairly similar, with the lagged REER coefficient always negative and significant at the one percent level, although the coefficient is biggest and most significant for the euro area.

In Table 3 we estimate a less parsimonious model of inflation as follows:

 (21)

In equation (21) the coefficient of the anchor country’s inflation rate is no longer constrained to be equal to one, and lagged inflation and lagged real exchange rate movements are added to the equation.  Because of the lagged dependent variable, the equation is estimated by OLS in every case, to avoid bias from the inclusion of fixed effects which, as in the case of Table 2, are insignificant for currency unions.

Table 3 shows the results of estimating equation (21).   For currency unions, both as a whole and individually, anchor currency inflation is highly significant with a coefficient exceeding but not very far from one, which implies that the Table 2 model is not unreasonable.  The lagged inflation coefficient is always positive and significant in almost every case, indicating some degree of inflation inertia. The lagged REER coefficients are similar to those in Table 2, with a point estimate of -0.0702 (and a t-statistic of-7.70) for currency unions, -0.0086 (-3.83) for other pegs and -0.0043 (-0.93) for floats.  For individual currency unions the lagged REER coefficient always has a slightly lower t-statistic than in Table 2, but it is still significant at 1 % in each case.[4]

To test whether currency unions are significantly different from other regimes, we estimate a pooled model with some or all of the coefficients allowed to vary with the exchange rate regime.  The results are shown in Table 4.  Column (1) of Table 4 shows a pooled version of the more parsimonious model shown in Table 2, estimated with country fixed effects.   The omitted regime category is a soft peg, so the regime coefficients in Table 4 are differences relative to a soft peg. The lagged real exchange rate coefficient is similar to that for soft pegs in Table 2 (-0.0126 compared with -0.116).  The coefficient of the lagged real exchange rate interacted with the currency union dummy is -0.0292, with a t-statistic of only -1.51, so the difference is not significant.  Partly this is because of the reduced size of the difference compared with Table 2, where the difference was -0.0542 (in Table 2 the insignificant country fixed effects were excluded for customs unions, whereas they are included in Table 4).  But it is also true that the higher standard error of the estimated parameter in Table 4 makes a big difference: if the standard error were the same as in the first column of Table 2, the coefficient of -0.0292 would have a t-statistic of -3.29, and would be significant at the one percent level.

Column (2) of Table 4 is a pooled version of Table 3, with the coefficients of the lagged real effective exchange rate and of the change in the anchor currency’s real exchange rate allowed to vary across exchange rate regimes, as in Column (1).  The baseline results are very similar to those shown in Table 3 for soft pegs, and this time the coefficient of the lagged real exchange rate interacted with the currency union dummy is -0.0464 and highly significant, with a t-statistic of -5.03.  Nevertheless the difference between currency unions and soft pegs is still smaller in the pooled model than in Table 3, where the difference in the lagged real effective exchange rate coefficients was greater than -0.06.

The models estimated so far assume that the explanatory variables are exogenous.  This seems reasonable: they are either predetermined, because they relate to the previous period, or if not they are anchor-currency variables, and therefore relate to a much larger economy (the United States or the Euro Area).   Thus the usual type of endogeneity, a feedback effect from the dependent variable to the independent variables, should not be a major concern here. A different sort of endogeneity issue is that countries with lower price rigidity might self-select into currency unions, so that the estimated parameters overstate the effect of a given country joining a customs union.  In this case it is not that the parameters are subject to simultaneous equation bias, but that their interpretation depends on how currency union (CU) membership is determined.  If any difference between currency unions and other regimes is the effect of self-selection, then there is no evidence that currency unions change agents’ behaviour, as the model above predicts.

The question is how to test for self-selection.  To address this issue, we exploit the fact that currency unions are geographically concentrated.   If we assume that this signals regional differences in price rigidity, then non-CU members in the same region as a CU should have lower price rigidity than other non-CU countries.  This can be tested by separating out non-CU countries in the same region as a CU from other non-CU countries.  If the self-selection argument is correct, the difference should be significant (although non-CU countries in the same region as a CU may still have greater price rigidity than CU members, which might explain why they have not joined the CU). Accordingly, in Columns (3) and (4) of Table 4, we add another regime category to those used in the regressions of Columns (1) and (2): non-CU members in Europe, West Africa or the Caribbean (defined as Z = 1). Adding these variables does not improve the fit, and in Column (4) in particular they are highly insignificant, although slightly less so in Column (3).  Thus, according to this test at least, there is no strong evidence that our results are driven by a self-selection effect.

Table 5 looks at nominal exchange rate adjustment relative to the US dollar for soft pegs and floats, with the same explanatory variables as used for relative price adjustment in Table 2.  The sample is confined to non-European countries, since European countries are more likely to be pegged to the euro, and also to observations with inflation in the range -10 % to +10 %. The estimated equation is:

                                         (22)

where NUS is the bilateral exchange rate against the U.S. dollar (U.S. dollars per unit of country i’s currency).

For floats, the coefficient of the lagged REER in Table 5 is -0.164, with a t-statistic of -5.02, indicating significant exchange rate adjustment against the US dollar.  For soft pegs, the coefficient is smaller, at -0.0577, and just significant at 10%, with a t-statistic of -1.71.  Nevertheless the coefficient is more than five times as large as the equivalent coefficient for relative price adjustment for soft pegs in Table 2, which suggests that soft pegs rely mainly on exchange rate rather than price adjustment.

A common explanation for slow price adjustment is the existence of nominal rigidities, and particularly resistance to falls in prices and wages.   If rigidities are particularly strong in a downward direction, then adjustment should be slower for negative shocks, which require price falls, than for positive ones.  This suggests that nominal rigidities should be reflected in asymmetries in the adjustment process.  To test this, the square of the lagged real effective exchange rate is added to equation (17).  If nominal rigidities are significant, then adjustment should be slower (implying a less negative coefficient) when the lagged REER is high, so the coefficient of the square of the lagged REER should be positive.

Table 6 shows the results for all currency unions combined, and then for each one separately.  For all currency unions together, the coefficient of the square of the lagged REER has the expected positive sign, but it has a t-statistic of only 1.11.  Of the individual currency unions, only the ECCA has a positive sign for this coefficient, which is just significant at the 5 % level.  For the Eurozone (where the results with country fixed effects are shown), the coefficient of the squared lagged REER is significant at the 1 % level but negative, in contradiction to the nominal rigidities hypothesis.

Table 7 repeats Table 2 using relative GDP deflators rather than consumer prices, because GDP deflators cover a wider range of goods and include exports rather than imports.  The point estimate of the lagged REER coefficient for all currency unions combined is even slightly larger than in Table 2, at -0.0801, but it is significant at only 10 %, because the root mean square error of the regression is 0.0554, compared with 0.0201 in Table 2.  This reflects the volatility of GDP deflators amongst commodity exporters in the CFA zone.  The lagged REER coefficient for other pegs is rather larger and more significant than in Table 2, at -0.0337, with a t-statistic of -4.65.    The individual currency unions present a very varied picture.  For the euro area, the lagged REER coefficient is 40 % larger than in Table 2.  This is consistent with Tressel and Wang’s (2014) finding that different measures produce different estimates of the degree of REER adjustment in the euro area.  For the ECCA, however, the lagged REER coefficient is unexpectedly positive, and for the CFA it is strongly negative, at -0.1965, but not statistically significant.

 

 

6.      CONCLUSIONS

Floats rely exclusively on nominal exchange rate adjustment to correct real exchange rate misalignments.  In the empirical analysis here, soft pegs are like paler versions of floats, with slower nominal exchange rate adjustment and negligible relative price adjustment.  In currency unions, where nominal exchange rate adjustment is ruled out, relative price adjustment is statistically significant but still quite slow in economic terms, with a ten percent higher real exchange rate being associated with a consumer price differential of less than one percent in the subsequent year.  For the euro area, the estimated rate of adjustment is faster using GDP deflators rather than consumer prices.  The role of nominal rigidities remains unclear, because the evidence does not demonstrate that adjustment is any slower in the case of negative shocks.

Our theoretical model predicts that agents change their behaviour because of the exchange-rate commitment of a currency union.  An alternative explanation of our results is that only countries with low price rigidity join currency unions.  It is difficult to rule out this possibility, but we have attempted to test it by assuming regional differences in price rigidity, exploiting the fact that currency union membership is geographically concentrated.  Our results suggest that the regional effect is small.

 

REFERENCES

Barro, R.J. and D.J. Gordon (1983) Rules, discretion and reputation in a model of monetary policy, Journal of Monetary Economics 12, 101-121

 

Bleaney, M.F., S. Bougheas and I. Skamnelos (2008) A model of the interactions between banking crises and currency crises, Journal of International Money and Finance 27, 695-706

 

Bleaney, M.F. and M. Francisco (2005), Exchange rate regimes and inflation – only hard pegs make a difference, Canadian Journal of Economics 38, 1453-1471

 

Bleaney, M.F. and M. Tian (2017) Measuring exchange rate flexibility by regression methods, Oxford Economic Papers 69, 301-319

Broda, C. (2004) Terms of trade and exchange rate regimes in developing countries, Journal of International Economics 117, 379-408

 

Edwards, S. and E. Levy-Yeyati (2005) Flexible exchange rates as shock absorbers, European Economic Review 49, 2079-2105

 

Eichengreen, B, N. Jung, S. Moch and A. Mody (2014) The Eurozone crisis: phoenix miracle or lost decade? Journal of Macroeconomics 39, 288-308

 

Frankel, J.A. and A.K. Rose (1997) Is EMU more justifiable ex post than ex ante? European Economic Review 41, 753-760

 

Ghosh, A.R., J.D. Ostry and M.S. Qureshi (2015) Exchange rate management and crisis susceptibility: a reassessment, IMF Economic Review 63, 238-276

 

Gibson, H.D., T. Palivos and G.S. Tavlas (2014) The crisis in the euro area: an overview, Journal of Macroeconomics 39, 233-239

 

Honkapohja, S.  (2014) The euro area crisis: a view from the north, Journal of Macroeconomics 39, 260-271

 

Jaumotte, F. and P. Sodsriwiboon (2010) Current account imbalances in the southern euro area, Working Paper no. 10/139,  International Monetary Fund

 

Mundell, R.A. (1961) A theory of optimum currency areas, American Economic Review 51, 657-665

 

Obstfeld, M. (1996) Models of currency crises with self-fulfilling features, European Economic Review 40, 1037-1048

 

Obstfeld, M. and K. Rogoff (1996) Foundations of International Macroeconomics, London: MIT Press

 

Organisation for Economic Cooperation and Development [OECD] (2016) Economic Survey of Ireland in 2016, Paris: OECD.

 

Shambaugh, J. (2012) The euro’s three crises, Brookings Papers on Economic Activity, Spring, 157-231

 

Tressel, T. and S. Wang (2014) Rebalancing in the euro area and cyclicality of current account adjustments, Working Paper no. 14/130,  International Monetary Fund

 

Whelan, K. (2014) Ireland’s economic crisis: the good, the bad and the ugly, Journal of Macroeconomics 39, 424-440


 

 

Appendix.  Countries in the Sample

 

 

Table A1. Sample of Countries and Groups (listed by 2012 exchange rate regime)

Currency Unions (37)

Euro area (17)

Austria, Belgium, France, Germany, Italy, Luxembourg, Netherlands, Finland, Greece (since 2001), Ireland, Malta (since 2008), Portugal, Spain, Cyprus (since 2008), Slovak Republic (since 2009), Estonia, Slovenia (since 2007);

CFA (14)

Cameroon, Central African Republic, Chad, Republic of Congo, Benin, Equatorial Guinea, Gabon, Guinea-Bissau, Côte d’Ivoire, Mali, Niger, Senegal, Togo, Burkina Faso;

East Caribbean (6)

Antigua and Barbuda, Dominica, Grenada, St. Kitts and Nevis, St. Lucia, St. Vincent and the Grenadines;

Soft Pegs (92)

Albania, Angola, Anguilla, Argentina, Aruba, Republic of Azerbaijan, Bahamas, Bahrain, Barbados, Belize, Bhutan, Bolivia, Bosnia and Herzegovina, Botswana, Brunei Darussalam, Bulgaria, Burundi, Cabo Verde, Cambodia, P.R. China: Hong Kong, P.R. China: Macao, P.R. China: Mainland, Comoros, Democratic Republic of Congo, Costa Rica, Croatia, Curaçao and St Maarten, Denmark, Djibouti, Dominican Republic, Egypt, El Salvador, Eritrea, Ethiopia, Fiji, Georgia, Guinea, Guyana, Haiti, Honduras, Indonesia, Iran, Iraq, Jamaica, Jordan, Kazakhstan, Kiribati, Kuwait, Lao People’s Democratic Republic, Latvia, Lebanon, Lesotho, Libya, Lithuania, Macedonia, Malawi, Maldives, Micronesia, Montenegro, Montserrat, Morocco, Myanmar, Namibia, Nicaragua, Nigeria, Oman, Pakistan, Panama, Papua New Guinea, Peru, Qatar, Rwanda,  São Tomé and Príncipe, Samoa, Saudi Arabia, Sierra Leone, Solomon Islands, Sudan, Suriname, Swaziland, Tajikistan, Tanzania, Thailand, Tonga, Trinidad and Tobago, Tunisia, Ukraine, United Arab Emirates, Vanuatu, Venezuela, Vietnam, Republic of Yemen;

Floats (52)

Afghanistan, Algeria, Armenia, Australia, Bangladesh, Belarus, Brazil, Canada, Chile, Colombia, Czech Republic, Gambia, Ghana, Guatemala, Hungary, Iceland, India, Israel, Japan, Kenya, Republic of Korea, Kyrgyz Republic, Liberia, Madagascar, Malaysia, Mauritania, Mauritius, Mexico, Moldova, Mongolia, Mozambique, Nepal, New Zealand, Norway, Paraguay, Philippines, Poland, Romania, Russian Federation, Republic of Serbia, Seychelles, Singapore, South Africa, Sri Lanka, Sweden, Switzerland, Turkey, Uganda, United Kingdom, United States, Uruguay, Zambia.


 


Table 2.   Lagged Real Effective Exchange Rates and Consumer Price Inflation Relative to the Anchor Currency

 

 

Exchange rate regime

Individual currency union

 

Currency union

Other peg

Float

CFA

Euro area

ECCA

 

 

 

 

 

 

 

Estimation method:

OLS

FE

FE

OLS

OLS

OLS

Constant

0.306***

(7.46)

 

 

0.299***

(3.55)

0.444***

(7.93)

0.279***

(3.53)

DRA

 

-0.0138

(-0.93)

-0.0338**

(-2.53)

-0.0112

(-0.60)

-0.0242

(-0.69)

-0.0329***
(-2.31)

0.0129

(0.50)

R(-1)

-0.0658***

(-7.42)

-0.0116*

(-1.73)

-0.0038

(-0.79)

-0.0642***

(-3.46)

-0.0963***

(-7.89)

-0.0599***

(-3.54)

 

 

 

 

 

 

 

Sample size

602

1447

798

190

226

186

No. countries

37

136

103

14

17

6

RMSE

0.0201

0.0243

0.0203

0.0296

0.0093

0.0175

p-value of country fixed effects

0.850

0.000***

0.000***

0.769

0.098*

0.796

Notes.  Dependent variable is change in log of consumer price index of member country relative to the anchor currency (DP – DPA). DRA: change in log of real effective exchange rate of anchor currency.  R:  log of real effective exchange rate of member country.  Figures in parentheses are heteroscadisticity-robust t-statistics. *, **, ***: significantly different from  zero at the 10, 5 and 1 % levels. Annual data 1980-2014 (1999-2014 for CFA and Euro area). For regimes other than currency unions, the sample excludes countries with consumer price inflation of more than 10 % or less than -10 %, and the anchor currency is the US dollar.  Regime classifications: Bleaney and Tian (2016).

 


 

Table 3.   A Less Restrictive Model

 

 

Exchange rate regime

Individual currency union

 

Currency union

Other peg

Float

CFA

Euro area

ECCA

 

 

 

 

 

 

 

Estimation method:

OLS

OLS

OLS

OLS

OLS

OLS

Constant

0.320***

(7.64)

0.0590***

(5.15)

0.0496**

(2.58)

0.215**

(2.46)

0.336***

(4.83)

0.279***

(3.41)

DPA

1.143***

(12.88)

0.589***

(9.23)

0.353***

(3.21)

1.511***

(3.83)

1.218***

(13.3)

0.959***

(9.09)

DRA

 

-0.0326**

(-2.06)

-0.0317**

(-2.30)

-0.0150

(-0.81)

-0.0180

(-0.48)

-0.0521***
(-4.77)

-0.0424*

(-1.68)

R(-1)

-0.0702***

(-7.70)

-0.0096***

(-4.04)

-0.0056

(-1.29)

-0.0487***

(-2.64)

-0.0750***

(-4.98)

-0.0607***

(-3.45)

DP(-1)

0.100*

(1.81)

0.212***

(2.88)

0.212***

(3.99)

0.146

(1.43)

0.233***

(2.97)

0.135**

(2.40)

DR(-1)

-0.0268

(-1.06)

-0.0326**

(-2.19)

-0.0075

(-0.86)

-0.1207**

(-2.12)

0.0291

(1.47)

0.0117

(0.31)

 

 

 

 

 

 

 

Sample size

595

1417

786

189

226

180

No. countries

37

136

103

14

17

6

RMSE

0.0197

0.0267

0.0245

0.0294

0.0086

0.0162

R-squared

0.2865

0.2188

0.2103

0.1838

0.6621

0.3782

Notes. Dependent variable is change in log of consumer price index of member country (DP).  DRA: change in log of real effective exchange rate of anchor currency.  R:  log of real effective exchange rate of member country. DPA: change in log of consumer price index of anchor currency.  See Notes to Table 2.


 

Table 4.  A pooled model

 

 

(1)

(2)

(3)

(4)

Estimation method:

FE

OLS

FE

OLS

Constant

 

0.0589*** (5.31)

 

0.0576*** (5.01)

DPA

 

0.621*** (12.8)

 

0.628*** (12.8)

DRA

-0.314** (-2.41)

-0.0350*** (-2.72)

-0.0261* (-1.73)

-0.0348** (-2.42)

R(-1)

-0.0126** (-2.16)

-0.0097*** (-4.05)

-0.0115* (-1.83)

-0.0095*** (-3.87)

DP(-1)

 

0.203*** (4.23)

 

0.204*** (4.24)

DR(-1)

 

-0.0170** (-1.96)

 

-0.0172*** (-1.98)

Currency union dummy (CU)

0.129 (1.45)

0.204*** (4.79)

0.133 (1.49)

0.206*** (4.82)

CU*DRA

0.0114 (0.62)

-0.0067 (-0.35)

0.0061 (0.31)

-0.0070 (-0.36)

CU*R(-1)

-0.0292 (-1.51)

-0.0464*** (-5.03)

-0.0303 (-1.56)

-0.0467*** (-5.06)

Float dummy (FL)

-0.0654** (-2.58)

-0.0135 (-0.54)

-0.0701** (-2.53)

-0.0126 (-0.47)

FL*DRA

0.0199 (0.93)

0.0273 (1.17)

0.0216 (1.01)

0.0274 (1.17)

FL*R(-1)

0.0149** (2.72)

0.0037 (0.50)

0.0159*** (2.70)

0.0035 (0.61)

Dummy for non-CU countries in region with CU  (Z)

 

 

 

0.0061 (0.24)

Z*DRA

 

 

-0.0243 (-1.21)

-0.0010 (-0.05)

Z*R(-1)

 

 

-0.0103 (-0.98)

-0.0010 (-0.18)

Sample size

2840

2795

2840

2795

No. countries

167

 

167

 

RMSE

0.0224

0.0248

0.0224

0.0248

R-squared

 

0.266

 

0.266

Notes. Dependent variable is change in log of consumer price index of member country (DP).  DRA: change in log of real effective exchange rate of anchor currency.  R:  log of real effective exchange rate of member country. DPA: change in log of consumer price index of anchor currency.  Z: dummy =1 for non-CU countries in Europe, West Africa and the Caribbean, and =0 otherwise.  See Notes to Table 2.

 


 

 

 

Table 5.  Nominal Exchange Rate Adjustment in Other Pegs and Floats

 

 

Exchange rate regime

 

Other pegs

Floats

 

 

 

DRA

 

-0.548***

(-12.6)

-1.00***

(-11.51)

R(-1)

-0.0566*

(-1.77)

-0.120***

(-3.44)

 

 

 

Sample size

1630

847

No. countries

136

111

RMSE

0.0624

0.0883

Notes. Country fixed effects included.  Dependent variable is change in the bilateral nominal exchange rate relative  to the US dollar. DRA: change in log of real effective exchange rate of anchor currency.  R:  log of real effective exchange rate of member country.  Figures in parentheses are heteroscedasticity-robust t-statistics. *, **, ***: significantly different from  zero at the 10, 5 and 1 % levels. Annual data 1980-2014.  The sample excludes countries with consumer price inflation of more than 10 % or less than -10 %, and the anchor currency is the US dollar.  Regime classifications: Bleaney and Tian (2016).


 

Table 6. Testing for Nominal Rigidities in Currency Unions

 

 

All

CFA

Euro area

ECCA

 

 

 

 

 

Estimation method:

OLS

OLS

FE

OLS

Constant

1.03

(1.57)

1.13

(0.76)

 

6.09**

(2.04)

DRA

-0.125

(-0.85)

-0.0210

(-0.58)

-0.0353***

(-3.01)

0.0205

(0.80)

R(-1)

-0.384

(-1.34)

-0.442

(-0.65)

3.64***

(3.14)

-2.53**

(-2.01)

R(-1) squared

0.0350

(1.11)

0.0427

(-0.55)

-0.409***

(-3.21)

0.264**

(1.97)

 

 

 

 

 

Sample size

602

190

226

186

No. countries

37

14

17

6

RMSE

0.0201

0.0297

0.0089

0.0173

R-squared

0.0881

0.0460

0.1813

0.1072

p-value of country fixed effects

0.832

0.779

0.015**

0.529

Notes. Dependent variable is change in log of consumer price index of member country relative to the anchor currency (DP – DPA). DRA: change in log of real effective exchange rate of anchor currency.  R:  log of real effective exchange rate of member country.  Figures in parentheses are heteroscedasticity-robust t-statistics. *, **, ***: significantly different from  zero at the 10, 5 and 1 % levels.  Annual data 1980-2014 (1999-2014 for CFA and Euro area). Regime classifications: Bleaney and Tian (2014).

 


 

Table 7

Using GDP Deflators instead of Consumer Price Inflation

 

 

Exchange rate regime

Individual currency union

 

Currency union

Other peg

Float

CFA

Euro area

ECCA

 

 

 

 

 

 

 

Estimation method:

OLS

FE

FE

OLS

FE

OLS

Constant

0.381*

(1.82)

 

 

0.922

(1.51)

 

-0.179**

(-2.02)

DRA

 

-0.205***

(-3.85)

-0.085***

(-4.69)

-0.018

(-0.91)

-0.505***

(-3.72)

-0.0642***
(-3.85)

-0.0689

(-1.50)

R(-1)

-0.0801*

(-1.77)

-0.0353***

(-4.96)

-0.0202**

(-2.12)

-0.1965

(-1.48)

-0.1346***

(-7.24)

0.0396**

(2.09)

 

 

 

 

 

 

 

Sample size

622

1625

830

192

226

186

No. countries

37

147

110

14

17

6

RMSE

0.0554

0.0325

0.0244

0.0893

0.0131

0.0297

p-value of country fixed effects

0.3020

0.000***

0.000***

0.807

0.000***

0.194

Notes. Dependent variable is change in log of the GDP deflator of member country relative to the anchor currency. DRA: change in log of real effective exchange rate of anchor currency.  R:  log of real effective exchange rate of member country.  Figures in parentheses are heteroscedasticity-robust t-statistics. *, **, ***: significantly different from  zero at the 10, 5 and 1 % levels. Annual data 1980-2014 (1999-2014 for CFA and Euro area). For regimes other than currency unions, the sample excludes countries with consumer price inflation of more than 10 % or less than -10 %, and the anchor currency is the US dollar.  Regime classifications: Bleaney and Tian (2016).

 



[1] For a model of the interaction between banking and currency crises, see Bleaney et al. (2008).

[2] This classification is chosen because of its data coverage.

[3] Data before 1999 are not used for the CFA zone because of the large devaluation of the CFA franc in 1994.

[4] If we ignore the bias issue and include country fixed effects where the test statistic shows them to be significant, the coefficients of the lagged REER are fairly similar: -0.00676 for other pegs and +0.00287 for floats.

 

网站导航  |  建议反馈  |  联系我们
铁道党校   版权所有   京ICP备14015019号  公安机关备案号
地址:北京市海淀区北太平庄路27号   邮编:100088   电子邮件:xgzx@tddx.cn
您是本站第 10082251 位访问者