1
U
NIVERSITY OF
A
MSTERDAM
The influence of exchange rate movements on current account balances in the EMU
Abstract: This study investigates the influence of exchange rate movements on the adjustment of the current account balance of the original EMU countries in the period 1980-2014. In particular, it is researched whether entering a currency area led to the loss of a potential adjustment mechanism for the CA balance. Besides, it is researched if the introduction of the common currency had effect on the relationship between exchange rate movements and the current account balance. Panel data regressions
are performed to obtain the results. In contrary with previous literature, there is no evidence found for the loss of a potential adjustment mechanism for the current account. In addition, there is neither evidence found for a potential adjustment mechanism to develop after the common currency has been
introduced. Although, this process might be on the longer term. This initiates further research on a longer time period of the common currency analysed.
Key words: Current Account, Exchange Rates, Optimum Currency Areas
JEL classification: F32, F33
Bachelor thesis Economics and Business Specialization Economics and Finance
Freek Smit, 10541977 Supervisor: Gabriele Ciminelli
Period:
2
Statement of Originality
This document is written by Freek Smit who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Contents
1. Introduction ... 4
2. Literature review ... 7
2.1 Economic grounds for a European common currency according to the OCA-theorem: ... 7
2.2 Monetary policy and the current account balance... 9
2.3 Effect endogeneity of OCA-criteria on the loss of monetary autonomy. ... 11
3. Research design ... 12
3.1 Data description ... 13
3.2 Methodology ... 17
3.3 Hypotheses ... 18
4. Results ... 19
4.1 Panel data regression analysis ... 19
4.2 Robustness checks ... 24
5. Conclusion ... 24
6. Reference list ... 25
4
1. Introduction
The Maastricht Treaty among European countries in 1992 led to the introduction of the Euro in 1999. Since this agreement there is a discussion about the grounds of the common
currency. The discussion addresses whether the reason for the common currency is based on economic or political grounds. The European Economic Commission outlined the economic benefits and costs of the common currency in 1990 (Emerson, 1992). The impression is given that the common currency is overall beneficial from an economic point of view. However, Feldstein (1998) states that European countries did not agree to implement the common currency based on economic grounds. According to Feldstein (1998), the Euro is implemented on political grounds. Countries abandoning their currency is a sign that part of their
sovereignty is handed over to the EMU. Handing over sovereignty improves the political grounds for a central government in Europe and it improves the unity among member countries.
Many commentators viewed the Euro as an economic success until the worldwide economic crisis in 2007. However, the following Euro-sovereign debt crisis in the so-called PIIGS (Portugal, Ireland, Italy, Greece, Spain) countries raised again questions about the economic impact of the common currency. The EMU countries have implemented the
common currency for a considerable amount of time now, this makes assessment of the effect of the common currency on economic indicators possible.
Prior to the Euro-sovereign debt crisis the PIIGS countries faced strong deficits on their current account (CA) balance relative to the other EMU countries. CA balances of the PIIGS countries developed to deficits since the introduction of the common currency while the other countries in the EMU, considered as the core countries, did not have deficits. Figure 1a shows the CA balance developments of the PIIGS or periphery countries and 1b shows this of the core EMU countries: Austria, Belgium, Germany, Finland, France and the Netherlands. The deficits on the CA balance and the Euro-debt crisis in the PIIGS countries can therefore not be considered as unrelated.
However, the consequences of the introduction of a common currency on the CA balance heterogeneity between the core and periphery EMU countries is unclear. A potential adjustment mechanism for the CA balance was lost by handing over monetary autonomy to the EMU. In this research it is analysed whether there was indeed an adjustment mechanism lost for the EMU countries. Two panel data regressions expose the relationship between ER movements and the CA balance. One for the core countries and one for the periphery
5
Figure 1a: Current account as % of GDP.
6
countries in the EMU. The two separate regressions are performed to expose if there was any difference between the core and periphery countries which might have contributed to the CA balance divergence. In contrast to literature based expectations there is no significant
relationship found between the CA balance and ER movements for both groups. This result suggests that no adjustment mechanism was given up. For the CA balance there is no reason to believe the common currency was not based on economic grounds. Still, the relationship between the ER and the CA balance might change through developments caused by the common currency. Although, in the analysis over the first 15 years of the common currency
Figure 1b: Current account as % of GDP.
7
there is no sign found for that.
In the next section the literature review is presented. The literature review consists of three parts and covers the consequences of the introduction of a common currency in general and specifically for the CA balance. In addition, developments after the introduction of the common currency are discussed. The third section contains the research outline which consists of the dataset description, the methodology and the hypotheses. In the fourth section the results are presented and interpreted. Also econometrical issues are discussed. Finally, a conclusion will be presented.
2. Literature review
The first part describes the economic consequences of entering a currency union according to the OCA-theorem. This part will highlight the loss of monetary autonomy as a consequence of entering a currency area. The second part describes the consequences of the loss of monetary autonomy for the CA balance by exposing the relationship between ER movements and CA balance adjustments. In the final section it is considered whether the endogeneity of criteria for an OCA has effect on the relationship between ER movements and CA balance
adjustments.
2.1 Economic grounds for a European common currency according to the OCA-theorem. A currency union is beneficial from economic perspective, if the benefits of the common currency outweigh the costs. Mundell (1961) introduced the theory of Optimum Currency Areas (OCA) to describe these benefits and losses of a common currency. The benefits are the loss of trade barriers and the loss of exchange rate risk. These factors provide economic integration due to lower costs and risk for trade. The costs of implementing a monetary union are represented by the disadvantages of giving up monetary autonomy. In a monetary union individual countries do not have the autonomy to conduct monetary policy individually. Countries hand over the sovereignty to conduct monetary policy to a central institution. That central institution serves the interest of all member countries in the union. Monetary policy is used to achieve macro-economic goals like price stability, low unemployment levels and sustainable economic growth. Therefore, the macro-economic goals the whole monetary union faces compared to those of individual countries determines the consequences of giving up this monetary autonomy. A centralized monetary policy might compensate the loss of
8
monetary autonomy if the needs and goals of the use of monetary policy are similar among the members of the monetary union. However, asymmetric economic shocks among members diverges the goals of using monetary policy. That is why asymmetric shocks should be
countered away by other adjustment mechanism than monetary policy. Mundell (1961) provides labour and capital mobility across regions within the currency area and wage flexibility as adjustment mechanisms to counter those asymmetric shocks.
Over the years, several authors added new criteria to the OCA-framework to assess the desirability of a monetary union. The added assessment criteria for an OCA are openness of trade (McKinnon, 1963), fiscal transfers and the diversification of the economic structure (Kenen, 1969). Improvement in those criteria lead to a reduction of the downsides of a monetary union. The criteria of Mundell, McKinnon and Kenen forms the base of the OCA theory. Desirability for a common currency is a complex issue to determine due to interaction between those OCA-criteria and the economic goals of a country.
Bayoumi and Eichengreen (1997) assess the desirability to introduce a common currency in European countries in an empirical model following the OCA-theorem, despite the complexity to assess if a common currency is desirable for a specific country. They analyse the core criteria for an OCA to determine whether a common currency is desirable for European countries. The OCA-criteria for each country are compared to Germany, because Bayoumi and Eichengreen view Germany as the core member of the European currency union. The analysis covers the period from 1980 until 1995. The benefits of a common currency with Germany should outweigh the costs for an OCA. Bayoumi and Eichengreen provide an index which shows to what extent a common currency is beneficial for the European countries. This OCA-index divides the European countries in three groups of desirability for a common currency according to the OCA-theorem. Countries who meet the OCA-criteria, countries who are developing to meet the OCA-criteria and countries who do not show development towards meeting the OCA-criteria. France, Finland, Greece, Italy, Portugal and Spain belong to the two groups of countries who did not (yet) meet the OCA-criteria in 1995 according to the OCA-index of Bayoumi and Eichengreen. Those countries did agree in 1992 to enter the currency union by signing the Maastricht treaty. Those countries entered the currency union despite not meeting the OCA-criteria, suggests not all countries had economic reasons to do so.
9
2.2 Monetary policy and the current account balance
As mentioned in the first part, countries use monetary policy to reach macro-economic goals like price stability, low unemployment and sustainable economic growth. However, the overall goal is to improve the total welfare of a country. To carry out monetary policy countries influence variables such as interest rates, monetary aggregates and exchange rates to. Changes in these variables have effects to several economic variables to reach their macro-economic goals. According to Pillbeam (2013), the CA picks up changes in the exchange rate. As a result, monetary policy can be used to influence the CA balance. A currency union has the consequence that countries do not have autonomy to conduct monetary policy anymore. Thus, the common currency has the consequence of giving up a potential adjustment
mechanism for the CA balance. Therefore, this section describes the role and the relationship with ER movementsof the CA balance in the economic analysis of a country.
The CA balance consists of the import and export of goods and services, the in- and outflow of interest payments, dividends and profits to other countries and the unilateral payments and receipts. It is part of a country’s balance of payments, which is the report off all economic transactions of the country with the rest of the world. The CA reflects the difference between domestic output and domestic spending of a country. A CA balance surplus implies that a country generates more income than it spends (Pillbeam, 2013). Countries with CA surpluses are considered to be net savers whereas countries with CA deficits are net
borrowers. Therefore, the CA balance reflects the domestic saving and investment behaviour of a country relative to other countries.
Debelle and Faruqee (1996) describes the role of the CA balance as a buffer to internal disturbances in output and demand. It functions as a shock absorber to cash flow or output changes resulting in the smoothing of consumption, which leads to maximization of the welfare-level of a country. The use of the CA balance to influence the welfare-level is an argument to incorporate the CA balance into the economic analyses of open countries, such as the EMU countries.
Pillbeam (2013) discusses the effect of ER movements on the CA balance. Starting point of the analysis is the elasticity approach. The elasticity approach describes the impact of ER movements on imports and exports. Exchange rate movements have two effects according to the elasticity approach. Firstly, the price effect which means that the value of trade on the CA balance will change after ER movements because the value of the underlying assets are different. An increase in the value of the domestic currency corresponds to an appreciation of
10
the ER, while a decrease in the value of the domestic currency corresponds to a depreciation. An appreciation of the ER increases the value of domestic exports and decreases the value of foreign imports. Therefore, the domestic spending decreases relatively to the domestic income and the CA balance increases. An increase of the CA balance after an appreciation implies that the CA and the ER have a positive relationship according to the price effect.
However, the second effect after ER movements of the elasticity approach, the volume effect, states that economic agents change their behaviour due to relative price changes (Pillbeam, 2013). As a depreciation implies a lower value of the domestic currency the domestic prices decrease relatively. Lower relative prices encourage the volume of exports and reduces the volume of imports since domestic products are relatively cheaper. As a
consequence, domestic spending decreases relative to domestic income. Thus, a depreciation improves the CA balance suggesting a negative relationship according to the volume effect. The negative price effect and the positive volume effect after a depreciation makes the net effect of ER movements on the CA balance ambiguous. The net effect depends upon whether the price or volume effect dominates. The Marshall-Lerner condition describes whether the price or volume effect dominates. According to the Marshall-Lerner condition, a depreciation leads to an increase of the CA balance if the sum of the foreign elasticity of demand for exports and the home country elasticity of demand for imports is greater than unity. If the sum of those elasticities is lower than unity a devaluation deteriorates the CA balance (Pillbeam, 2013).
In contrast to the Marshall-Lerner condition, the J-curve effect states that a
depreciation lead to a deterioration of the CA balance initially due to the price effect. The volume effect is not immediately observed, as consumers and producers need time to adjust to the new market circumstances (Pillbeam, 2013). This adds a time-dimension to the analysis of the effect of ER movements on the CA. Variation over time is in line with empirical results where longer-term demand elasticities are bigger than the short-term demand elasticities. Demand elasticities that increase over time suggests that in the longer-term the Marshall-Lerner condition might hold.
In addition to the elasticity approach, Pillbeam (2013) describes the effect of ER movements on the CA balance according to the absorption approach. The absorption approach states that changes in imports and exports, as result of the price and volume effect, have influence on domestic income and domestic spending. If the domestic income grows relatively to the domestic spending the CA balance improves. Several effects influence the domestic income and domestic spending as a result of ER movements. However, the net
11
effect remains ambiguous due to different speed of adjustment over time and relative effect on the determinants of the domestic income and domestic spending. Nevertheless, the absorption approach adds a dynamic approach to the analysis of the effect of ER movements on the CA balance. This dynamic approach states that as several variables are influenced by the ER movements, other variables are influenced by those developments.
Arghyrou and Chortareas (2008) support that ER movements do influence CA balance dynamics. In their empirical research they find a long-run significant negative relationship between real ER movements and CA dynamics in EMU countries in the period from 1990 until 2005. The negative relationship is in line with the Marshall-Lerner condition and the J-curve effect (Pillbeam, 2013). The negative relationship between real ER movements and CA dynamics suggests that giving up monetary autonomy by entering the EMU indeed led to the loss of an adjustment mechanism for the CA balance.
2.3 Effect endogeneity of OCA-criteria on the loss of monetary autonomy.
Bayoumi and Eichengreen (1997) state that economic integration (measured by the increase in bilateral trade) is one of the factors to determine whether a common currency is desirable from an economic perspective. According to them the introduction of a common currency might lead to economic integration. As a common currency implies stable exchange rates which encourage bilateral trade. These findings suggest that monetary and economic integration are endogenous and constitute a “virtuous, self-reinforcing circle”. So, at first a common currency area might not be desirable for an area, but after the common currency is introduced the area might converge to an OCA in which the common currency is desirable. Frankel and Rose (1998) expand this endogeneity approach. They state that business cycle correlation is also endogenous besides trade integration suggested by Bayoumi and Eichengreen (1997). Both factors influence the desirability of a common currency, but the common currency also influence business cycle correlation and trade integration. However, the influence the common currency has on the business cycle correlation is unclear.
According to Frankel and Rose, the common currency might influence production of a country into two different directions. The first direction states that production becomes more similar between countries after the introduction of a common currency. Development towards more similar production will happen if demand predominates shocks or if intra-industry trade accounts for most of a country’s trade. If production becomes more similar among countries the business cycle will converge. When convergence occurs, the economic shocks that
12
countries face will be less asymmetric. Reduction in asymmetric economic shocks improves the desirability of a common currency according to the OCA-theorem (Mundell, 1961). However, the alternative direction states that production of countries will be more specialized if the focus is on comparative advantages. Production specialisation will lead to a more different composition of the business cycle between countries. The countries are therefore sensitive to different industry specific shocks. Business cycle divergence will lead to more asymmetric economic shocks, making a common currency less desirable.
European countries introduced the common currency more than 15 years ago. The time the common currency is present in the EMU makes this endogeneity of trade integration and business cycle correlation an interesting concept to analyse. This endogeneity approach can also be considered for CA balance dynamics. Reduction in asymmetric shocks is beneficial for the single monetary policy to compensate for the loss of monetary autonomy (Frankel and Rose, 1998). Although, the influence of ER movements on the CA balance might change after the introduction of the common currency due to different composition of the trade and business cycle.
Arghyrou and Chortareas (2008) found a relationship between real ER movements and in- or decreases of the CA balance among EMU countries between 1990 and 2005.
Furthermore, Arghyrou and Chortareas observed that the countries facing real ER
depreciations/appreciations are the same countries facing CA balance in-/decreases in the post-EMU period. This observation implies that real ER movements have a similar effect on the CA balance among the EMU countries. The heterogeneity in CA developments indicates therefore structural economic differences between EMU countries according to Arghyrou and Chortareas (2008). This heterogeneity in CA developments suggests that the common
currency led to business cycle divergence instead of convergence described by Frankel and Rose (1998). As a consequence, the central monetary policy is not sufficient to compensate the loss of monetary autonomy for the CA balance due to the asymmetric developments. Although this result should be interpreted carefully due to the short period (6 years) that Arghyrou and Chortareas (2008) researched, while endogeneity might be a longer-term process.
3. Research design
This section covers the structure of the research. First the outline of the panel data regression analysis is discussed in the data description section. This section describes the time period and
13
the countries covered in the panel data regressions. Also descriptions, sources, reasons to include and descriptive statistics of the variables included in the regression are presented. The methodology section discusses the regression technique, forms of the variables and
econometrical issues. Finally, the hypotheses section describes the research questions and the expected answers.
3.1 Data description
This research aims at exposing the effect of ER movements on the CA balance in EMU countries. Therefore, the countries that are included in the dataset are the original EMU countries Austria, Belgium, Finland, France, Germany, the Netherlands, Ireland, Italy, Portugal and Spain. Luxembourg is excluded due to the insignificant role this country has in international economics. Also Greece is included, since this country introduced the common currency 3 years after the start of the EMU in 2002. Other member countries of the EMU are not analysed due to short period of membership in the EMU. The EMU countries are divided into two groups. The first group consists of the core countries in the EMU. These are Austria, Belgium, Finland, France, Germany and the Netherlands. The second group consists of the PIIGS countries in the EMU. This group is defined as the periphery countries and it consists of Greece, Ireland, Italy, Portugal and Spain. Heterogeneity in developments of the CA balance between the two groups of countries is the reason to divide the countries. Goal of the distinction between core and periphery countries is to expose if the heterogeneity is caused by different developments of the CA balance after ER movements.
The time period that is covered in the regression analysis is the period from 1980 until 2014. The period 1980 until 1999 is covered to analyse the relationship between ER
movements and the CA balance in the upcoming period towards the introduction of the Euro. In this period each country had their own currency. This period is analysed to expose if entering the currency union led to the loss of an adjustment mechanism for the CA balance. The period from 1999 until 2014 is included to analyse whether the relation between the CA balance and the ER changed due to endogeneity of OCA-criteria.
The CA balance is included in the regression analysis relative to the domestic GDP to compare CA dynamics across countries. The CA balance and all variables included are taken annually. A broad description and sources of all variables can be found in table 1.
14
Table 1: dataset description
Variables Source* Variable Description
CA WDI, CIA Current Account to domestic GDP ratio.
REER WDI Real Effective ER (index rate with 2010 as base year = 100). INF** WDI CPI index inflation relative to global mean.
GDP IFS Real GDP growth
SOD WDI Stage of Development: domestic GDP per capita
measured relative to GDP of the USA.
SAV WDI Gross domestic savings relative to domestic GDP.
GOV*** OECD Government debt level relative to domestic GDP.
DEPY WDI Youth dependency ratio (relative to global mean).
Population under 15 / population between 15 and 64.
DEPO WDI Old dependency ratio (relative global mean).
Population over 64 / population between 15 and 64.
*Data sources used to construct the corresponding variables: WDI: World Development Index by the World Bank, IFS: International Financial Statistics by the IMF, OECD: dataset of the Organization for Economic Cooperation and Development, CIA: The World Factbook of the Central Intelligence Agency.
** For INF data from 1980 to 1999 is missing for Germany *** For GOV data from 2010 to 2014 is missing for all countries
The ER is included in the form of the real effective exchange rate (REER). The REER is the value of the nominal ER adjusted for inflation and weighted against an index of foreign currencies. The weights of each foreign currency in the REER index depends on the relative level of trade balance of a foreign country’s currency against each country within the index. Movements of the REER purely reflect changes in exchange rate weighted against the trading partners of a country. Making the REER an indicator what the relative price effect is of an appreciation or a depreciation (Pillbeam, 2013). An increase in the REER corresponds to an appreciation.
However, the aim of this research is to expose whether the handing over of monetary autonomy led to giving up an adjustment mechanism for the CA balance. Therefore, relative inflation is included as a control variable to cancel out the possible effect of inflation
differentials between countries on the effect of REER movements on the CA. Inflation is included relative to the world average to expose these inflation differentials (Chinn and
15
Prasad, 2003). As the effect of inflation differentials between countries is cancelled out, only the effect of nominal ER movements on the relative price can be interpreted out of the REER coefficients in the regression analysis.
Furthermore, several control variables are included in the regression analysis to cancel out other factors affecting the development of the CA balance. First of all, real GDP growth is included. Real GDP growth reflects the growth of domestic income adjusted for inflation. Therefore, real GDP growth might have influence on the CA balance (Debelle and Faruqee, 1996).
Secondly, the stage of development(SOD) is included. The SOD of a country is determined as a country's GDP per capita relative to the GDP per capita of the United States (Chinn and Prasad, 2003). Following Roldos (1996), relative low developed countries invest to improve their stage of development. As a consequence, the country spends more than their domestic income and will face CA deficits. On the contrary, developed countries have CA surpluses by paying off the imported capital and export capital to less developed countries. This implies a positive relationship between the CA balance and the stage of development. Also the gross domestic savings as a part of the GDP level are included. The gross domestic savings are calculated as GDP minus the total consumption. This variable reflects the saving behaviour and indicates the difference between domestic income and domestic spending. Increases in the level of savings indicate a greater gap between domestic income and domestic spending. This implies a positive relationship between the savings and the CA balance.
Also the government debt level is included as a control variable. The government debt level is included as a percentage of the GDP. According to Obstfeld & Rogoff (1998)
government deficits are used to redistribute income from future to present generations. Redistributing income from future to present generations induces CA deficits since the government spends more than it generates income. An increase in the debt level induces a reduction in the CA balance. Suggesting a negative relationship between the government budget balance and the CA.
Finally relativedemographic developments are included. Following Chinn and Prasad (2003) the domestic spending pattern relative to the domestic income is influenced by the structure of the population. Displayed by the relative dependency ratios for the young (𝐷𝐸𝑃𝑌𝑖,𝑡) and the old (𝐷𝐸𝑃𝑂𝑖,𝑡) population. Young dependency consists of the people
younger than 15 relative to the working age population. The old dependency consists that of people older than 64. Both ratios are taken relative to the global mean of those dependency
16
ratios. The dependency ratios are taken relatively to the global mean because only relative differences are important for cross-country differences in savings (Chinn and Prasad, 2003).
In table 2 the mean and standard deviation of the dependent, independent and control variables are presented for both groups of countries. These descriptive statistics are presented to give some insight of the relative behaviour of these variables between those groups. The CA balance of the core countries is higher and less deviating relative to the CA balance of the periphery countries. The REER is divided in two time periods via a time corresponding dummy. REER LCU corresponds to the period 1980 to 1999, where each country had their local currency. REER EUR corresponds to the period 1999 to 2014, where all countries had a common currency. Table 4 shows that the REER of the core country and the periphery country groups converges and has a lower standard deviation in the second period. The common currency among the two groups of countries explains this development.
Table 2: descriptive statistics
Core countries Periphery countries
Mean St. Dev. Mean St. Dev.
CA 1.80% 3.23 -2.64% 3.92 REER LCU 104.76 9.16 90.72 10.67 REER EUR 99.73 3.41 96.76 5.98 GDP 1.96% 2.18 2.17% 3.15 SOD 0.83 0.14 0.54 0.22 SAV 25.27% 3.03 21.60% 6.75 GOV 53.50% 25.71 69.31% 30.01 DEPY 0.54 0.07 0.57 0.12 DEPO 2.14 0.20 2.09 0.33 INF 0.44 0.24 0.88 0.68
Table 3 presents a decomposition of the variance of the dependent, independent and control variables. The decomposition of the variance divides the standard deviation to variation over entities and to variation over time for this panel data set. Several variables show more variation over the time dimension, while other variables show more variation in the cross-country dimension. There are also differences between the two groups of countries, though for most variables the relative importance of the time and country dimension is similar. Main observation is that many of the variables vary in both the time and the entity dimension.
17
Table 4: Decomposition standard error into time and cross
country differences (in percentage points)
Core countries
Periphery
countries
Across countries Across time Across countries Across time
CA 51.84 48.16 47.74 52.26 REER LCU 83.05 16.95 81.44 18.56 REER EUR 42.08 57.92 27.95 72.05 GDP 5.07 94.93 44.18 55.82 SOD 23.60 76.40 77.24 22.76 SAV 78.63 21.37 82.10 17.90 GOV 91.84 8.16 94.95 5.05 DEPY 77.44 22.56 93.99 6.01 DEPO 81.90 18.10 89.32 10.68 INF 18.66 81.34 42.47 57.53 3.2 Methodology
In order to expose the relationship between the CA and the ER a regression for both the core countries group and the periphery countries group is performed. Table 4 indicates that the variables included are both varying in the time dimension and the entity dimension.
Therefore, fixed effect panel data regressions are performed for both groups. The panel data regression analysis is performed in the following model:
𝐶𝐴𝑖,𝑡 = 𝛽0+ 𝛽1𝐶𝐴𝑖,𝑡−1+ 𝛽2𝑅𝐸𝐸𝑅𝑖,𝑡−1∗ 𝐿𝐶𝑈𝑡+ 𝛽3𝑅𝐸𝐸𝑅𝑖,𝑡−1∗ 𝐸𝑈𝑅𝑡+ 𝛽4𝐺𝐷𝑃𝑖,𝑡
+ 𝛽5𝑆𝑂𝐷𝑖,𝑡+ 𝛽6𝑆𝐴𝑉𝑖,𝑡+ 𝛽7𝐺𝑂𝑉𝑖,𝑡+ 𝛽8𝐷𝐸𝑃𝑌𝑖,𝑡+ 𝛽9𝐷𝐸𝑃𝑂𝑖,𝑡+ 𝛽10𝐼𝑁𝐹𝑖,𝑡 + 𝛾1𝑌𝐸𝐴𝑅2𝑡+ ⋯ + 𝛾𝑇−1𝑌𝐸𝐴𝑅𝑇𝑡+ ∝𝑖 +𝑢𝑖,𝑡
The dependent variable in the regressions is the 𝐶𝐴𝑖,𝑡 corresponding to the CA to GDP
of a country. The CA balance of the year before is included as independent variable (𝐶𝐴𝑖,𝑡−1).
This lagged variable functions as a precautionary measure for first-order autocorrelation of the dependent variable.
𝑅𝐸𝐸𝑅𝑖,𝑡−1 is included as explanatory variable. Instead of the time-corresponding variable of the REER, the lagged value (𝑅𝐸𝐸𝑅𝑖,𝑡−1) is included. The lagged value is included to prevent possible simultaneous causality bias between the CA and the REER. Firstly, the lagged REER is interacted with the dummy 𝐿𝐶𝑈𝑡. This dummy corresponds to the period
18
from 1980 to 1999. In this period each country in the EMU had an individual domestic
currency. The interaction variable between the REER and the dummy 𝐸𝑈𝑅𝑡 covers the period with a common currency in the EMU from 1999 to 2014. Except for Greece where the LCU period is from 1980 to 2002 and the EUR period starts from 2002. Another reason to include the lagged variable of the REER is that CA balance has an adjustment period after relative price changes, because the behaviour of economic agents does not change immediately after price changes.
𝐺𝐷𝑃𝑖,𝑡, 𝑆𝑂𝐷𝑖,𝑡, 𝑆𝐴𝑉𝑖,𝑡, 𝐺𝑂𝑉𝑖,𝑡, 𝐷𝐸𝑃𝑌𝑖,𝑡, 𝐷𝐸𝑃𝑂𝑖,𝑡 and 𝐼𝑁𝐹𝑖,𝑡 are all factors that
influence domestic income, domestic spending or relative prices. Therefore, those factors are included in the model as control variables. The control variables are added to prevent biases due to omitted variables (Stock and Watson, 2015).
The fixed effect regression model controls for variables that are constant over time but differ across entities (Stock and Watson, 2015). This entity effect is captured in the variable ∝𝑖. Though, the fixed effects model does not control for effects that vary over time but are constant over entities. Therefore, time dummies are added for all minus one year
(𝛾1𝑌𝐸𝐴𝑅2𝑡+ ⋯ + 𝛾𝑇−1𝑌𝐸𝐴𝑅𝑇𝑡) to avoid estimation biases caused by time-specific effects. One time dummy is excluded to avoid multicollinearity.
3.3 Hypotheses
Goal of the panel data regressions is to expose the effect of ER movements on the CA balance in the EMU countries. The relation between ER movements and the CA is exposed to analyse if handing over monetary autonomy in EMU countries had the consequence of giving up an adjustment mechanism for the CA balance. According to Arghyrou and Chortareas (2008) there is a relationship between the real ER and CA developments in the EMU countries. However, it remains unclear if this relationship exists if the real aspect of the ER is cancelled out by taking the inflation differentials into the model. If there exists a relationship between the REER and CA balances after the real aspect of the REER is cancelled out there was indeed an adjustment mechanism for the CA balance giving up by entering the EMU. According to the Marshall-Lerner condition a negative relationship between ER movements and the CA balance exists, if the sum of import and export elasticities exceed unity. Though,
19
the J-curve effect and Absorption approach describe time effects and dynamic developments as potential reasons for the Marshall-Lerner condition not to hold (Pillbeam, 2013). Still, the first hypothesis that will be tested is the following:
Hypothesis 1: β (REER*LCU) < 0
Next tot that the consequences of the introduction of the Euro for the effect of ER movements on the CA balance is considered to analyse the divergence of the CA balances among EMU countries. The effect of ER movements on the CA balance might change after the common currency was introduced, due to endogeneity of OCA-criteria. However, the heterogeneity in CA development between the core and periphery countries was caused by asymmetric developments in relative prices, reflected by the REER, instead of different effects of REER movements on the CA balances in the period 1990 to 2005 according to Arghyrou and
Chortareas (2008). Though, the endogeneity process might be a longer term process. Since the period of the common currency that is analysed is 15 years it is tested if there are signs for this endogeneity process. Therefore, the coefficients of the period before and after the common currency are tested via a Wald-test for significant changes (Stock and Watson, 2015). However, the direction whereto the effect might move remains unclear since the direction of the endogeneity is ambiguous.
Hypothesis 2: β (REER* EUR) ≠ β (REER* LCU)
4. Results
In this section the results of the panel data regressions are analysed. The results are presented, interpreted and explained. In the robustness check potential estimation biases are discussed. 4.1 Panel data regression analysis
Table 4 presents the results of the fixed effects panel data regressions of the core countries group. In the regressions the influence of the REER on the CA balance is researched. The time analysed is split up in two periods. REER LCU corresponds to the period before the common currency, while REER EUR corresponds to the period after the common currency
20
was introduced. To check for potential misalignments of the relationship between the
dependent variable CA and the explanatory variable lag of the REER due to omitted variables several control variables are added. The control variables are included in the analysis to check for their impact on the relationship between the dependent (CA) and explanatory variable (REER). Regression 6 is therefore the most appropriate, since in that regression the effects of all influencing control variables on the relationship between the REER and the CA are cancelled out.
Table 4: regression results core countries with CA to GDP as dependent
variable 1 2 3 4 5 6 Constant -4.027 -2.272 0.718 -1.172 -2.921 -21.794 (7.412) (7.449) (7.414) (7.076) (8.716) (12.73) Lag CA 0.857*** 0.874*** 0.887*** 0.779*** 0.69*** 0.530*** (0.041) (0.042) (0.042) (0.047) (0.06) (0.083)
Lag REER LCU -0.036* -0.044** -0.006 -0.049* -0.023 0.081
(0.019) (0.019) (0.024) (0.025) (0.032) (0.100)
Lag REER EUR 0.051 0.034 0.051 0.002 -0.023 -0.011
(0.074) (0.074) (0.073) (0.071) (0.085) (0.034) GDP -0.111 -0.087 -0.239*** -0.147* -0.013 (0.067) (0.067) (0.073) (0.085) (0.121) SOD -5.406** -5.938*** -4.039 -1.170 (2.101) (2.005) (2.983) (3.319) SAV 0.311*** 0.400*** 0.456*** (0.075) (0.112) (0.125) GOV 0.019 0.025 (0.016) (0.016) DEPY -9.512 (8.255) DEPO 4.064** (1.775) INF 0.395 (0.890) Significant time 4 4 3 6 8 2 Dummies R² 0.87 0.87 0.87 0.88 0.82 0.86 Number of 204 204 204 204 160 149 Observations
Note: symbols *, ** and *** indicate statistical significance at 10%, 5% and 1% level. Standard errors are presented in the brackets below.
21
As presented in table 4, there is no significant relationship between the CA and the lag of the REER in the LCU period of the core countries in regression 6. The first hypothesis, a negative relationship between the REER and the CA balance, is therefore rejected for the core countries group.
The J-curve effect and the absorption approach are potential explanations for the Marshall-Lerner condition not to hold and thus the first hypothesis rejected. The J-curve effect states that the time dimension used in the analysis has influence on the relationship between the ER and the CA. The absorption approach describes developments in other economic variables as a reason for the Marshall-Lerner condition not to hold. The dynamic effects of this approach cause the CA balance to develop differently. However, since several economic developments are included as control variable this explanation seems unlikely. Another potential explanation for rejection of the first hypothesis in contrast to the findings of Arghyrou and Chortareas (2008), is cancelling out the real part of the REER movements by including inflation differentials. However, even if the inflation differentials are not included as control variable (regression 5) there is no significant relationship observed between the CA balance and the REER LCU. Including inflation differential is therefore not a potential explanation.
Though, there is no relationship between REER movements and the CA balance for the core countries on this time dimension. This observation suggests that entering the currency union did not led to giving up an adjustment mechanism for the CA balance. Thus, losing monetary autonomy had no consequences for the core countries to influence their CA balances. Therefore, it cannot be stated that the decision of introducing a common currency was not based on economic grounds for this economic indicator the CA balance.
The second hypothesis is tested via a Wald-test. The Wald-test exposes if the
coefficient of the lag REER EUR variable has significantly changed relative to the coefficient of the lag REER LCU variable. According to the Wald test, output can be found in table 6 in the appendix, the relationship of the REER EUR and the CA balance relative to the
relationship of the REER LCU and the CA balance did not change significantly. Therefore, the second hypothesis is rejected for the core countries.
This suggests that the endogeneity of OCA-criteria have no effect on the relationship between REER movements and the CA balance. However, it is also possible that the
endogeneity had not yet any effect on the relationship, since this process might be on the longer-term. As there is no sign of endogeneity in the relationship between the REER and the CA balance, the interpretation and consequences of introducing a common currency does not
22
change.
Interpretation of the control variables GDP, SOD, SAV, GOV, DEPY, DEPO, INF is not the aim of this research. Besides, coefficients of control variables are in general biased and thus have no causal interpretation according to Stock and Watson (2015). Still, they function to control for over- or underestimated estimation of the explanatory variables. The R-squared indicates the explained variance of the dependent variable(CA) by the independent variables. The R-squared of the regressions indicates that large part of the variance is explained in the regressions. The significant time dummies indicate omitted effects that are constant across countries but differ over time.
In table 5 the regressions result for the periphery countries group are presented. The structure of the panel data regression on the periphery countries is similar to the panel data regression of the core countries group. Again regression 6 is the most appropriate, since in that regression effects of other variables on the relationship between the REER and the CA is countered away.
As presented in table 5, there is no statistical evidence for a relationship between the CA balance and the lag of REER movements of the LCU period in regression 6. For the periphery countries group the first hypothesis stating a negative relationship between the REER and the CA balance is also rejected.
Potential reasons for this rejection are similar to those of the core countries, namely the time dimension explained by the J-curve effect. Also the dynamic effect explained by the absorption approach can be interpreted comparable to the results of the core countries. Although, it should be noted that less control variables have impact on the relationship between the REER LCU and the CA balance. Before the inflation differentials are cancelled out, see regression 5, there is a negative relationship between the REER LCU and the CA balance observed in contrast to the core countries group. This result is in line with the observation of Arghyrou and Chortareas (2008).
The consequences for the discussion are the same for the periphery countries as for the core countries. Rejection of the first hypothesis implicates that no adjustment mechanism for the CA balance is given up by handing over monetary autonomy. Also for the periphery countries group a Wald test is performed to check if the REER EUR changed relative to the REER LCU. Again it turns out that the REER EUR did not change significantly, as can be found in table 6 in the appendix. Potential reasons are similar to those of the core countries group and so this observation does not change the consequences of entering the currency union (yet).
23
Table 5: regression results periphery countries with CA to GDP as dependent variable
1 2 3 4 5 6
Constant 6.658** 7.61*** 6.602** 6.024** 6.948 2.536
(2.764) (2.775) (2.805) (2.966) (5.13) (5.925)
Lag CA 0.692*** 0.715*** 0.690*** 0.68*** 0.566*** 0.612***
(0.061) (0.061) (0.062) (0.066) (0.087) (0.092)
Lag REER LCU -0.051** -0.057** -0.051** -0.048** -0.082*** -0.064
(0.023) (0.023) (0.023) (0.024) (0.029) (0.040)
Lag REER EUR -0.058** -0.066** -0.059** -0.056** -0.098*** -0.047
(0.027) (0.027) (0.027) (0.027) (0.033) (0.038) GDP -0.170** (0.070) SOD 0.173 (1.275) SAV 0.020 (0.034) GOV 0.013 0.018 (0.023) (0.023) DEPY 16.925** 18.774** (7.405) (7.470) DEPO -4.268* -4.584* (2.370) (2.369) INF 0.800 (0.547) Significant time 17 14 17 17 11 13 Dummies R² 0.85 0.84 0.85 0.86 0.57 0.56 Number of 170 169 170 170 138 138 Observations
Note: symbols *, ** and *** indicate statistical significance at 10%, 5% and 1% level. Standard errors are presented in the brackets below.
The control variables in table 5 are again not the aim to interpret, besides the output is biased according to Stock and Watson. The R-squared is lower for the periphery countries group than for the core countries group. Still, a great part of the variance in the CA balance is explained. There are more time dummies significant for the periphery countries than for the core countries, indicating that there were more time-specific effects in the periphery countries which seems appropriate due to the higher standard deviation of the variables included (see descriptive statistics)
24
4.2 Robustness checks
Besides theoretical issues on the interpretation of the results also econometrical issues might influence the results. In this section several econometrical issues to the interpretation of the results are discussed.
Firstly, the presence of serial correlation forms a threat for the internal validity of the results. According to Stock and Watson (2015) the time dimension in the panel data set makes the variables vulnerable for serial correlation. In the regressions the lagged variable of the CA is included to prevent first-order serial correlation. Though, Wooldridge’s test for
autocorrelation (Drukker, 2003) is performed to test if higher-order serial correlation is present. Output of Wooldridge’s test for the panel data regressions for both groups of countries are presented in table 7 which can be found in the appendix. The results are all strongly significant for both groups of countries and all models. Suggesting serial correlation still forms a bias in this panel data regression. Even when adding second or third lags to the model, serial correlation was still present. Other solutions to this bias is unfortunately beyond the scope of the bachelor’s thesis. The results should therefore be interpreted carefully.
The regression technique used for the panel data regression is the time fixed effect model. Assumption of the fixed panel data regression is that individual effects of countries are allowed to be correlated with the dependent variables. The Hausman test is performed to judge whether the fixed effects model is appropriate for the panel data regressions performed. The Hausman test analyses if the fixed effects model or the random effects model is more efficient. Output of the Hausman test for the panel data regression are presented in table 8 which can be found in the appendix. The Hausman test shows no significance results. The random effects model would therefore be more efficient in estimating the relationship
between the CA and the independent variables. The random effects model is again beyond the scope of this research.
5. Conclusion
Since the Maastricht Treaty in 1992 there is a debate about the grounds of the common currency. The discussion questions if the Euro is introduced based on economic or political arguments (Feldstein, 1998). The Euro-debt crisis raised again questions if the common currency is beneficial from economic perspective. In this research the CA balance, an economic indicator, is researched. CA balances in the PIIGS countries deteriorated strongly
25
in comparison with the core countries of the EMU since the common currency was
introduced. This heterogeneity among EMU countries raises questions about the impact of the common currency on the CA balance developments. Therefore, the consequences of
introducing the common currency on the CA balance is analysed in this research. In the panel data regression analysis there is no significant relationship found between REER movements and the CA balances for both group of countries. Neither is there any sign (yet) for the endogeneity of the OCA-criteria to have influence on this relationship. This suggests that no potential adjustment mechanism for the CA balance is given up by entering the currency union. The heterogeneity of CA balances among EMU countries is not caused or could be prevented by the common currency. Therefore, for the CA balance it cannot be concluded that there were no economic grounds to implement the common currency neither has that
conclusion changed through the endogeneity process. Although the results should be
interpreted carefully due to several econometrical biases, time dimension used and potentially omitted economic developments which influences the relationship between REER movements and the CA balance.
For further research, it would be interesting to research a longer time period of the common currency to analyse whether there is on the longer term are signs of the endogeneity process having influence on the relationship between ER movements and the CA balance. Next to that, more advanced econometric techniques to counter threats for the internal validity of the panel data regressions are desirable.
6. Reference list
Arghyrou, M. G. and G. Chortareas, 2008. Current Account Imbalances and Real Exchange Rates in the Euro Area. Review of International Economics, Vol.16(4), pp.747-764.
Bayoumi, T. and B. Eichengreen, 1997. Ever Closer to Heaven? An
Optimum-Currency-Area Index for European Countries. European Economic Review, 41(3-5), pp. 761– 770.
Chinn, M. D. and E.S. Prasad, 2003. Medium-Term Determinants of Current Accounts in Industrial and Developing Countries: An Empirical Exploration. Journal of International Economics, Vol 59, pp. 47–76.
Debelle, G. and H. Faruqee, 1996. What determines the current account? A cross-sectional and panel approach. IMF Working Paper 96/ 58.
26
Drukker, D. M., 2003. Testing for serial correlation in linear panel-data models. Stata
Journal, 3(2), 168-177.
Emerson, M., 1992. One market, one money: an evaluation of the potential benefits
and costs of forming an economic and monetary union. Oxford University Press on Demand.
Feldstein, M., 1998. The political economy of the European economic and monetary union: political sources of an economic liability. Journal of Economic Perspective, Vol. 11, No. 8, pp. 23-42.
Frankel, J. and A. Rose, 1998. The Endogeneity of the Optimum Currency Area Criteria. The Economic Journal, Vol. 108, No. 449, pp. 1009-1025.
Freund, C., 2005. Current account adjustment in industrial countries. Journal of International Money and Finance, Vol.24(8), pp.1278-1298
Kenen, P., 1969, The theory of optimum currency areas: An eclectic view. in R.A. Mundell and A.K. Swoboda (Eds.), Monetary problems of the international economy
(University of Chicago Press, Chicago, IL), pp. 41-60.
McKinnon, R., 1963. Optimum currency areas. American Economic Review, Vol. 53, pp. 717-725.
Mundell, R., 1961. A Theory of Optimum Currency Areas. American Economic Review, 51(4), pp. 657-665.
Obstfeld, M. and Rogoff, K., 1998. Foundations of International Macroeconomics. MIT Press, Cambridge, MA.
Pillbeam, K., 2013. International Financ, 4th edition. Hampshire: Palgrave Macmillan. Chapter 3 -- pp…
Roldos, J., 1996. Human capital, borrowing constraints, and the stages of the balance of payments. Manuscript, IMF.
Stock, J.H., and M.W. Watson, 2015. Introduction to Econometrics, revised 3th edition. Essex: Pearson Education Limited.
27
7. Appendix
Table 6: Wald test for inequality REER*LCU and REER*EUR
F-value
Core countries 0.98
Periphery countries 2.23
Table 7: Wooldridge's test statistics for autocorrelation
1 2 3 4 5 6
Core countries 45.02*** 44.89*** 46.74*** 79.91*** 47.64*** 32.20***
1 2 3 4 5 6
Periphery countries 83.81*** 85.78*** 79.26*** 69.9*** 93.34*** 7.43** Note: symbol *** indicates statistical significance at 1% level
Table 8: Hausman test statistics for random versus fixed effects
1 2 3 4 5 6
Core countries 7.21 6.18 13.1 11.84 11.57 21.88
1 2 3 4 5 6
