Wage flexibility and export in a currency area

(1)

1

Ivan Peynirdzhiev

Student number: 6106226 peynirdzhiev@gmail.com

Master thesis

Universiteit van Amsterdam

Faculty of Economics and Business Study: Economics

Supervisor: Dr. D.J.M. Veestraeten

30.11.2015

Wage Flexibility and Export in a Currency Area

(2)

2

I. Introduction.

As the economic integration of the world is advancing further, even through the hurdles caused by the worldwide economic crisis, the question of the exchange rate regimes and monetary integration will continue to be of interest to policymakers and economists particularly in Europe where the Euro zone is an ongoing project. The success of the Euro zone has stakes in the success of the European project as a whole, but its outcome will also without doubt be looked as an example (whether good or bad is still to be seen) by other groups of countries in the world with similar aspirations for monetary integration. The theory of optimum currency areas provides a framework that allows not only for weighing the advantages and shortcomings of a proposed currency area, but also examining the required preconditions for a country to be a successful member of one. This study concentrates on the criterion claiming that a certain degree of wage flexibility is needed to address internal and external imbalances in the absence of flexible exchange rates and independent monetary policy. Using as a basis the gravity model of international trade the link between unit labour cost and export is studied as a method to examine the relation between wage flexibility and external imbalances.

II. The theory of optimum currency areas.

In the theory of optimum currency areas Mundell (Mundell 1961) investigates the benefits and shortcomings of flexible and fixed exchange rates and independent monetary policy in regions or countries based on their factor mobility. A currency area is an entity, within which the exchange rates are fixed, whether by commonly pegged currencies or by a single currency. By theorizing about currency area there is a need to explain what the concept of region is – what is meant by region is an economic entity that produces either just one tradable (what is the example of Mundell) or in the more realistic case tradables that are subject to the same type of demand and productivity shocks. The simplest example of the difference of flexible and fixed exchange rates is when the regions are considered to issue a national currency, i.e. they are sovereign countries and there is no factor mobility between them. In such a setting a demand shock that shifts the demand curve to the left can be dealt with an exchange rate devaluation that will reduce the real wage and ensure that competitiveness is regained. This example shows that floating exchange rates and independent monetary policy make sense in such environment, but in the real world the national economies are

(4)

4

not regionalized in this way. We have countries issuing a single currency with many different regions. For example if a country has two regions producing two different products, the negative demand shock for either of the regions will produce unemployment and external deficit for that region and the central bank cannot use expansive monetary policy to bring down unemployment without causing inflation in the other region. In this situation the exchange rate depreciation/appreciation and independent monetary policy fail to address the different effect of the shocks in one country – more than one instrument is needed to meet more than one goal. Thus the theory proves that if the world is comprised of countries (as currency issuers), which do not coincide with regions (in the sense that was explained) and do not have factor mobility between them, then the flexible exchange rate mechanism cannot prevent inflation and unemployment. Still if there is coordination between monetary policies of the regions, there is a trade-off between unemployment in one region and inflation in the other one that can be fine-tuned. But then the same trade-off can be achieved by a unified monetary policy in a single currency area, created by sovereign nation states.

Let’s now look at a setting, which allows for a currency area where factors of production are perfectly mobile. If the product market demand shifts from a product produced by one region to one produced by another region, the loss of employment in the first region is to be compensated by the increase in the other and given the mobile labour and capital, no unemployment or decrease in participation rates and inflation should occur in neither region. Consider now the same case, but with perfect wage flexibility (no employment is lost regardless how much the labour demand falls) instead of labour mobility. The product (pencils, for example) market demand shift will reduce the labour demand in one region and the wage flexibility will enable the producers to stay competitive. There will be a change in the distribution of welfare between the regions, but the employment level will be preserved. The welfare of the region producing pencils will decrease and the increase will be distributed by the regions buying pencils. This simple example obviously assumes no voluntarily quits occur due to the wage decrease. However neither perfect labour mobility nor perfect wage flexibility are realistic assumptions. So what are the challenges that face countries without perfect labour mobility and wage flexibility in a currency area? Here are further questions that shed light on the problem:

1) What role does the loss of the devaluation mechanism play during a demand shock?

It is a challenge for countries within a currency union to deal with the asymmetric shocks that affect regions in the country. Usually if a country suffers a recession because of demand slowdown, the nominal exchange rate will depreciate, and given that the inflation stays the same, this will be followed by a real exchange rate devaluation. This will make the products cheaper, boosting competitiveness. Losing this mechanism means regaining competitiveness trough nominal exchange rate is not possible. What are possible policy responses? If a country is relatively big enough so that the region affected with a negative labour demand shock could be stimulated out of recession at a little aggregate cost for the country’s fiscal stability, then this second best solution is going to suffice as a response to the negative shock. The first best solution for the country is a currency union-wide fiscal authority. A union-wide fiscal authority means there will be intra-union welfare transfer from the tax payers of the union to the affected region1_{. Here} by stimulus it is not meant necessarily only discretionary spending, but also the automatic adjustment mechanisms as lower tax revenue and bigger social spending. It has to be said that the

(5)

5

present situation in the Euro Zone where countries do have regions with varying production structures is to an extent better than if the countries consisted of regions with similar structures. So even though there is a historical specialization of the European countries dating far before the EMU, there is a significant degree of variation in the structures of their economies. But still countries do not face similar challenges: small countries have to fight a tougher battle to keep the balance between fiscal stability and a recovery. This is generally for two reasons: first because smaller economies usually have bigger trade to GDP ratio, and second, small countries have narrower specialization in industries and the relative part of a single industry to GDP is bigger than in a country with a wider specialization. This means that there is a bigger fiscal cost to the adjustment mechanism. Of course this argument is not valid in each individual comparison between a smaller and a bigger country, but there is a robust advantage for bigger countries (Alesina et all 2005).

2) What role does the loss of independent monetary policy play role?

The entrance in a currency union is most often connected with the giving up of the currency devaluation mechanism that can improve international competitiveness. The mechanism of country specific monetary policy is also lost. This can be seen as problem on two levels. First, obviously a different monetary policy is needed when there is a misalignment of the business cycle. Second, different labour market characteristics with various degrees of wage flexibility can influence the transmission of the monetary policy of the whole currency area. Chrisoffel and Linzert (2005) look at wage rigidities as a source of persistent inflation. The wage setting regime, corporate or competitive, is not the decisive factor, but bargaining power of workers and employment protection. The argument is that lower volatility in wages desired by workers slows the adjustment of production costs and can lead to persistence in inflation. Thus monetary policy should regard different wage flexibility effects on inflation.

Before delving further into the role of wage flexibility in a currency area, a further look at the concept of wage flexibility on a basic level is needed to understand it better.

III. What is wage flexibility?

1) How to define wage flexibility?

There is no universally accepted definition of wage flexibility in economic literature. Historically the term wage flexibility is more broadly used since the late 1970s when unemployment mated with inflation in Europe and the US is moving into the focus of economic research. Qualities attributed to movements of the price of labour are also termed wage rigidity, stickiness and moderation and each term has its own connotation. In classical and neo-classical economics the price of labour has the characteristics of all other prices and is solely governed by the forces of demand and supply. In the world of Walrasian general equilibrium wages move freely and long-term unemployment can only stem from the conscious choice of economic agents between work and leisure. First John Maynard Keynes introduces nominal wage rigidities with the claim that workers will be much less inclined to accept a reduction in money-wages (Keynes’ term for nominal wages) then a real wage reduction through increase in prices (Keynes 1936, Chapter 19,II). Wage moderation is the preferred term to describe wage reductions in European labour markets with the corporatist wage bargaining and describes more a policy measure than a market

(6)

6

characteristic. As mentioned above, wage flexibility is used as early as 1961 by Mundell as a condition for viability for entrance in an optimum currency area. In this context I will elaborate different definitions of wage flexibility.

What is wage flexibility? One way to define it is via the function that it serves. According to the optimum currency area theory, wage flexibility is such a quality of the economy that enables it to adjust to demand shocks of its output, so that it will return to internal and external balance. Wage flexibility thus influences inflation, unemployment and current account balance - this is why it is of interest by policy makers. It is appropriate then to inquire wage flexibility in regards to the purpose it serves to each one of them. In order to see what wage flexibility could mean as a mechanism, consider this term on a micro level. Is wage flexibility a firm’s ability to maintain the employment level for a lower output, resulting in lower hours worked (work rationing) or the ability to lower its wage costs per output, in response to lowering of the price of that output, without laying-off workers? Since work rationing is effectively resulting in unemployed hours that employees are willing to work at the given wage, but are unable to, under wage flexibility in the narrow sense is to be understood only the later. Lowering the price of its output without lay-offs means that wage flexibility is effective if the adjustments made are in price, not in output of physical entities, because lowering of the output volumes will make some workers redundant and lead to lay-off no matter what their wage is. This supposes that there is certain elasticity of product demand so that lowering the price during a product demand shock can maintain output volume at the same level. There is another possible look at wage flexibility: if we consider that the lowering of product demand will inevitably lead to lowering of output volume and thus to layoffs, wage flexibility is the firm’s ability to hire at a lower wage so that its product becomes more competitive and the firm returns to the previous level of output and employment. Lay-offs could be a result from lower earnings not due to lower output, but lower price of the output. The difference between the two cases is not of the reason for lowering of the wages, but that in the second case the process happens through the mediation of unemployment. Following from the above a possible definition of wage flexibility encompasses two mechanisms: first, as the elasticity of wages of current workers to output prices, and second, as elasticity of wages of future workers to unemployment. That means that again in the narrow sense wage flexibility is only the mechanism that does not increase the unemployment. Next, I am going to elaborate further on the elasticity mechanism of wage flexibility in the narrow sense.

Wage flexibility, as the elasticity of wages to output prices (this would be understood as the narrow definition of wage flexibility) is actually the extent and speed to which wages w align to marginal product of labour (MPL), because MPL equals the revenue from the marginal product sold, which in turn is the same as the price of all products – I assume the firm sets the same price for unit of a given product. This gives w=φMPL where φ is the wage flexibility. Kniesner and Goldmith (1987) describe this definition (“the more closely and quickly the nominal wage mimics

the variation in the value of labor's marginal product over the business cycle, the more one might say a simple neoclassical labor market underlies the data” (Kniesner and Goldsmith p. 1255), as

a measure of the explanatory capabilities of neoclassical economics modeling the labour market as an auction ruled by labour demand and supply. So if this definition of wage flexibility is given, does this mean that it is a measure of the extent to which individual firms and workers behave like

homo economicus? Maybe not, as this reflects an outcome on the labour market and this is a result

of many arrangements, as it will be shown below. This definition of wage flexibility as the elasticity of wages to output prices accentuates that a fluctuation in marginal productivity of labour is first to happen: for example, it changes in response to a exogenous shock, i.e. the change in

(7)

7

wages plays a catch up role. I propose that wage flexibility is to denote also the case where the change in wages is not a response to a change in marginal labour productivity, but to other factors, so that wages deviate from MPL. Whereby in this case the greater extent and speed that wages change the lower the wage flexibility. Defined like this, the term includes not only the process of wages returning to equilibrium (w=MPL), but also the departure from equilibrium. It is better to invent a new term to identify the different cases, but as they have the same consequences to unemployment and competitiveness I will use wage flexibility as describing both.

To see whether this mechanism will serve its purpose to preserve a given level of employment, let’s consider a simple model. Under perfect wage flexibility 1= w/MPL as elasticity of wages to output prices, all reductions in prices are transferred instantaneously to reductions in wages by the same percentage. As w=MPL is the condition for maximizing profit, this means that a profit-maximizing firm cannot increase its profit by laying off workers and thus will keep the same level of employment. In a monopolistic competition with positive profits this means that the net profit margin under perfect wage flexibility will remain the same even as the sales and profits will decrease. However, for this to be true, the cost of capital should be perfectly flexible too and this not a plausible assumption. All things being equal, cost of capital even has a negative elasticity to output prices, because a company with decreasing sales is considered more risky. This means that even under perfect wage flexibility net profit margin will fluctuate in response to output prices. I turn my attention to net profit margin, because even though maximizing profit (minimizing losses) can dictate to keep the current level of employment, diminishing profit margin could lead to closing of the business altogether. Therefore perfect wage flexibility regarded as perfect wage elasticity to output prices might not guarantee keeping the levels of employment for an entire economy if profit margins fall and capital cost is not flexible.

This narrow definition of wage flexibility measures the lowering of the cost without laying-off employees, so in effect it shows to what extent the existing contracts between employer and employee can be renegotiated. This means that it is a characteristic more of the firms themselves than a labour market characteristic. Even though the definition regarding wage flexibility as not occurring through unemployment means that at the firm level there are no lay-offs, the wage negotiations are held on a specific background that includes labour market conditions as the unemployment in the region and in the country, the wage level, inflation and other economic information.

Until this point wage flexibility was narrowly defined as the extent of the ability of firms to reduce the labour cost without resorting to lay-offs or lowering work hours, so that cyclical unemployment would not arise, but this is a very limiting scenario, in reality unemployment rises during economic slumps. Now let’s leave the micro level of the firm and consider a whole labour market. If we consider that unemployment inevitably does arise during a demand shock, be it due to insufficient wage flexibility, business closures or other reasons, then wage flexibility is the elasticity of wages to unemployment. This is the more common usage of the term (Teulings and Hartog 1998, Chapter 7 p.235) and I will call is a wide definition of wage flexibility. This is represented by this equation:

𝑢 = −𝜑𝑤 where u is unemployment, w – wages and φ is wage

flexibility. When is φ=∞ wages do not follow changes in unemployment whatsoever and φ=0 means perfect wage flexibility, where any increase in unemployment is resolved with a decrease in wages at the time of the occurrence of unemployment. So basically this reflects the neo-classical view that unemployment is only voluntary. The effect of a product demand shock on unemployment comes through the elasticity of job destruction (unemployment) to output costs, the elasticity of wages to unemployment and the elasticity of job creation i.e. unemployment (if

(8)

8

there is no labour mobility and the participation rate is constant) to wages. These are separate mechanisms and should be distinguished.

It is interesting to see what the relation between the narrow and wide definitions is. The narrow wage flexibility does not influence the wide one, but the elasticity of wages to unemployment affects the elasticity of wages to output prices, because the contract negotiations do take notice the labour market conditions. Thus it is evident that these are not narrow and wide definitions in the traditional sense where the wide encompasses the narrow one, but are rather distinct.

To summarize the effectiveness of wage flexibility, in the narrow definition as the elasticity of wages to output prices, in curbing unemployment lies on the elasticity of product demand – will the lowering of the prices of the product manage to keep the volumes being sold?. In the broader definition, as elasticity of wages to unemployment, it lies on the elasticity of job creation to wages.

The difference between these two concepts of wage flexibility can be illustrated with the following example. If the labour demand shock is caused by labour demand innovation like technology change that is expected to be permanent, then the wage flexibility will keep the employment, but at the cost of lower wages for whole period until workers quit and look for a better paid job. Inside a company this means a shared sacrifice between the different professions - all to accept a lower pay in order a few to keep their job. On a bigger scale in a country this means that a company rendered unnecessary by being inefficient will be subsided to operate at the cost of shared sacrifice of the whole country. The other scenario - if employment is cut at the cost of an increase in unemployment expenditures - has the advantage that the people will be in the labour market faster and can look for a better paid job right away, then if they would still be at work at lower wage. Depending on the lavishness of the unemployment benefits and that cost of qualification programs on one side and the inefficiency (the size of the needed subsidy and wage decreases) on the other side it can be calculated Disregarding the social costs it would seem that it is always better to let the adjustment go through unemployment than trough lower wages of existing contracts, because it shortens the adjustment time. Whatever are the policy preferences of the policymakers it is likely that during an adverse labour demand shock, both the unemployment will increase and because of this there will be a downward pressure on wages of the newly hired. Unemployment will return to its equilibrium state under the influence of outward migration of workers and job creation triggered by the lower wages. So the speed of return is dependent on both the elasticity of worker migration to wages and the elasticity of job creation to wages.

2) What influences wage flexibility?

What are the factors that make wages react differently to demand shocks and unemployment? These can be differentiated into two groups of factors or mechanisms at play – endogenous and exogenous. Using a HR management survey Jonas Agell and Helge Bennmarker (2003) study the endogenous reasons that hamper greater flexibility. Endogenous mechanisms are attributed to social norms or psychological characteristics and are not directly affected by government policies. Such mechanisms come by the employee’s and employer’s attitude towards fairness, gift exchange, money illusion, fear from retribution (negative reciprocity), motivation mechanisms, risk of increasing quits etc. transaction costs). Their results show significant wage rigidities (only 1.1% of wages were nominally cut during a severe and prolonged recession) due to these mechanisms and their interaction with exogenous mechanisms.

(9)

9

The exogenous mechanisms are labour market institutions like government regulations on employment and employee protection, unemployment insurance, long term contracting, trade unions and their behavior. The effect of institutions on wage flexibility is ambiguous. For example, difficulties in hiring and firing procedures, on one side give bargaining power to the worker thus prohibiting greater flexibility, but also they create incentives for employer to lower costs through wage cuts rather than lay-offs, because these have greater transaction costs. Unions also have ambiguous effects on wage flexibility. It depends mostly on what their goal is: higher wages or preventing unemployment; whether they want to protect only their members or to serve the whole work force. The outcome also depends on the quality of the relations between unions and employers. If each views the other as a credible partner and realizes the interdependence between the counterparts, a cooperative solution can emerge, with each side not abusing their momentous bargaining power.

Bargaining regime also plays role in the case where wages are decided upon on an industry level firms only adjust employment. This is shown by the difference between efficient Nash bargaining and right-to-manage bargaining. In Nash bargaining unions maximize utility, which includes wages and employment and in right-to-manage bargaining unions maximize only wages and employment is subsequently determined by the firms.

The size of the employer plays a significant effect on wage rigidity. The relation between downward wage flexibility and presumed profitability is considered to be much stronger in large firms than in small ones. One explanation is that lowering wages in large firms is seen to affect motivation more than in small ones. The proximity of executives and owners to employees in small firms makes the communication much easier and motivations drops are more easily addressed. The communication distance problem in large firms, can be resolved with the help of a union that is seen as credible by both workers and management. This leads to a possible role for unions in increasing wage flexibility.

Helmund and Zetterberg (1991) study the inter-sector wage differentials and find that where the wage setting is more decentralized there is much more correlation of wages and industry performance than in centralized wage setting environment. Such decentralized regimes are not represented well by the classical labor demand and supply model and the wage setting depends on the bargaining powers of the sides which include the sector profitability. The profits are distributed between the firms and workers on rent-sharing bases. This leads to the thought that there isn’t a sufficient inter-sector labour mobility and sector labour markets work as separate entities. I don’t think the supply-demand and bargaining models are incompatible. They come from a different perspective. The supply and demand model states that if the unemployment is above the natural rate, then the wages must be above the equilibrium, because there is excess supply of labour at that wage level. The pressure is then downward on the wages towards equilibrium. Whereas in the wage bargaining model the higher unemployment lessens the bargaining power of the laborers, suppressing the wages. The line between the good applicability of the two models is the degree of inter-sector mobility.

During the times of both high inflation and unemployment in the late 70’s there were calls to make the labour cost a prearranged proportion of the revenue. In such an institutionalized arrangement of flexible wages there is a performance bonuses component of salaries. This makes workers stakeholders in the economic activity because they directly bear the consequences of the economic performance of their employer, they are becoming entrepreneurs in that that they share the gains and losses, but as decision makers their power could vary between none and quite significant (Weitzman 1984).

(10)

10

3) Nominal and real wage flexibility

The factors influencing wage flexibility presented above all point to the difficulties of lowering nominal wages. How about real wages? Kniesner and Goldmith provide valid arguments for the transaction costs of changes of nominal wages, but they also equal them to the changes in real wages. Reducing nominal wages indeed has all the costs they ascribe to it, but pay-freeze and lower inflation adjustment usually have lower costs. This would mean a higher degree of wage flexibility during times of higher inflation. Easier real wage flexibility in an inflationary environment could be a positive development for economies that have to adjust their labour cost against its trade partners. This wage adjustment would lead to higher inflation differentials within the currency area, with the countries that are doing the adjustment having lower inflation. The issue with this is that the monetary authority has to tolerate higher inflation for the whole currency area. A similar argument for the link of real and nominal wage flexibility and growth can be made. While in a high growth environment reduction of nominal wages is not needed, and pay-freeze is enough to lower real wages, in periods of low growth nominal flexibility is required to lower real wages. Fehr and Goette (2002) find evidence that nominal rigidities are persistent during both low and high growth periods, causing persistent unemployment during periods of low growth.

IV. Wage flexibility in a currency area

Why rigid wages? When Mundell weighted the advantages and shortcomings of flexible exchange rates and a bigger currency area comprising of many regions, a key argument against a currency area is that people are not inclined to accept real wage decrease through nominal wage decrease, but they are less against a decrease through currency depreciation. Why would people have a different reaction on how their real income is decreased? There are a few factors that might explain this. Let’s regard decentralized wage bargaining where usually individuals don’t have sufficient information. First, as seen above there are numerous factors for both employees and employers to accept or make nominal wage cuts. Second, I think people view their contracted wage as something that they can stand up for, whereas the international currency markets are something totally out of their reach and the reach of their employers, so a devaluation of the currency is seen as an outside force having little to do with their nominal wages.

The empirical evidence of the role of wage flexibility for the US presented by Blanchard and Katz (1992), suggests that wages don’t play significant role in dealing with imbalances. They are not interested in the external imbalances and do not study the effect on interstate trade, but on employment levels and unemployment. After a labour demand shock, wages do fall, but this does not lead to the creation of new jobs. This illustrates the individual mechanisms used in the wide definition of wage flexibility in chapter III.1). The elasticity of wages to unemployment is high, but the elasticity of job creation to wages is low. This would lead to persistent lower wages and unemployment if there is no labour mobility and the participation rate is the same. In the US labour mobility is high and because of that there is not persistent unemployment, rather the workers are moving out of state changing permanently the levels of employment. Decressin and Fatas (1994) estimate the model for Europe. Labour mobility between European countries is limited due to multitudes of languages and cultures and other factors, and labour demand shocks are mainly resolved through changes in participation rates. Unemployment as in the American case returns to its natural state rather quickly.

(11)

11

V. Empirical Study

The main part of this paper is a study of the effect of wage flexibility on exports. This is done on the background of the Mundell theory of optimum currency areas, claiming that in a currency area wage flexibility is needed in order for countries to deal with internal and external imbalances caused by asymmetric demand shocks in the absence of flexible exchange rates and independent monetary policy. The current empirical study concentrates on the external imbalances. It seeks to answer the question whether and how much changes in wages have effect on the changes in exports. A positive answer would mean that flexible wages help competitiveness. If we follow the narrow definition of wage flexibility as the extent to which wages follow the marginal product of labour the relation of wage flexibility to export is in reality the price elasticity of the demand for their export. This can have implications for both existing and prospective members of a currency union. Existing members can overcome at least partly their external imbalances by lowering their unit labour cost. This could also have implications in the policies of governments of countries seeking to be a part of a currency union. By examining institutions they can find approaches that would stimulate higher wage flexibility. As seen in the literature review above, next to the labour market institutions, there can be many of the factors that influence the nominal flexibility of wages that are out the control of the government and are part of societal and business culture. Next the model is set up.

1) The Gravity model.

The model on which the study is based is the Gravity model of international trade. First set up by Jan Timbergen to explain the observable trade flows between countries, as the Ricardian theory of comparative advantages did not match the data of global trade. The regression of the model is

𝑃𝑋

_𝑖𝑗

= 𝛼 + 𝛽

₁

𝑌

_𝑖

+𝛽

₂

𝑌

_𝑗

+ +𝛽

₃

𝐷

_𝑖,𝑗

+ 𝛽

₄

𝐴

_𝑖𝑗

+ 𝜀

_𝑖𝑗

(1)

where 𝑃𝑋_𝑖𝑗 is the trade flow between countries i and j, 𝑌_𝑖 and 𝑌_𝑗

are the respective sizes of their

economies, 𝐷_𝑖,𝑗is the distance between them and 𝐴_𝑖𝑗 is any other factor that aids or hampers trade. So in essence this is a Newtonian model where the trade between two countries is determined by their size and the distance between them (Bergstrand 1985).

To construct the regression, which will be used to answer the research question, the following variables are used:

 Export.

Export is the dependant variable. In the original gravity model trade flow is used because the focus is total trade volume. When the focus is competitiveness many studies use current account balance. I use exports and not CAB for a couple of reasons. First, form a practical point of view there is not a readily available data for the current account balance of a country with each separate trading partner. Second, in the current account are included income and transfers that are not connected

(12)

12

with competiveness. Here of course there can be a critique that not using CAB does betray the optimum currency area theory, which is not concerned with competiveness per se, but with the external balances. However I do not see an economic logic of how competitiveness can affect these entities (especially for transfers, though one might have an argument about income). So this additional data that does not have explanation in the model muddles the results of the regression.

The independent variables that measure wage flexibility and other factors that can influence export as foreign direct investments (FDI) and credit.

 Unit labour cost.

Unit labour cost is the relation between the average cost of labour per unit of output. In essence ULC shows how much is spent on labour compensation for a unit of output in an economy. On a micro level ULC represents the cost of labour per item produced ULC=average labour cost/unit produced. So when comparing two firms that produce the same product and pay the same wages we mean that they have the same ULC. Now if the first firm lowers the wage (or the other raises it), this will mean that the first firm will have lower ULC. This would suggest that ULC is a good measure of wage flexibility. When aggregating the ULC on firm level to the ULC for the whole economy problems arise that lead to a critique of the use of ULC. The problem lies in the fact that one economy produces many different products and because of this an economy wide ULC uses GDP per worker as the unit of output. But not all of the firms output that is included in the GDP is in tradables, which are important for competitiveness. This is a valid critique if you use ULC directly as a measure of competitiveness. I do not have such a claim, I ask whether changes in wages explain part of the changes in export, i.e. contribute to competitiveness, but this is something that has to be kept in mind when analysing the results. The equation includes nominal unit labour cost as a measure of wage flexibility, which uses nominal wages, but real GDP. Eurostat defines it as NULC = (D1/EEM) / (B1GM/ETO) where D1 is compensation of employees in all industries at current prices, EEM is employees in all industries in persons, B1GM is gross domestic product at market prices in millions, chain-linked volumes with a reference year 2010 and ETO is total employment in all industries, in persons. It is easy to see that there is a problem with using ULC as an independent variable in a regression where export is the dependent variable. Export is a component of GDP, which in turn is in the numerator of ULC - the dependent variable influences the independent variable. Thankfully, using bilateral trade means that the share of export in GDP is much smaller than if using total export. Still this has to be considered in the result analysis. For the main regression equation I use the difference between the nominal unit labour cost indexes of each individual pair of countries - ∆𝑈𝐿𝐶_𝑖𝑗𝑡

 GDP

In line with the gravity model I use real gross domestic product as a measure of size of the economies

 FDI

Foreign direct investments are added to the model because they can have an effect on exports that can otherwise be attributed to wage flexibility. The basis for this is the theory of multinational

(13)

13

enterprise that claims that multinational enterprises have a number of advantages over local firms such as better knowledge of the international markets, better technology, lower production costs, better supply chain ties and others (Kutan and Vuksic 2007). These advantages can boost exports without having effect on wages.

 Credit

Bank credit to non-financial corporations is added as in the case of FDI’s because of its probable influence on exports. The rationale behind this is that payments due from export are usually less certain and take longer time. This might lead to liquidity problems for firms that are operating on foreign markets, hence the bigger need for credit (Buono and Formai 2014). Moreover during the financial crisis after 2008 a much bigger drop in international trade was observed, compared with the drop in GDP for many counties. Given the sharp tightening of finance during this period Chor and Manova argue that credit constraints are partly responsible for the disproportionate drop in trade (Chor and Manova 2012).

For dynamic gravity models with country pairs, such as this, Baltagi (Baltagi et al 2014) uses the logs of trade and GDP. Using the logs omits FDI from the regression, because there are negative values in the data, thus giving:

𝑙𝑛𝐸

_𝑖𝑗𝑡 = 𝛼 + 𝛽₁∆𝑈𝐿𝐶_𝑖𝑗𝑡 +𝛽₂𝑙𝑛𝐺𝐷𝑃_𝑖𝑡 + 𝛽₃𝑙𝑛𝐺𝐷𝑃_𝑗,𝑡+ 𝛽₄𝑙𝑛𝐹𝐷𝐼_𝑖𝑡+ 𝛽₅𝑙𝑛𝐶_𝑖𝑡 +

𝜀

_𝑖𝑗𝑡 (2)

The main contribution of this empirical study is the use of the gravity model for estimation of competitiveness. Not only using bilateral trade data brings the possibility of having many more observations, but also the data itself is much more specific. If only data for total export and a unit labour cost relative to an average ULC is used, the results can be misleading because exports to different countries have different weight in the total export of each economy. This means that when regressing using a relative ULC, the results will be valid only when each country exports the same amount to each of its trade partners. Using bilateral trade data also means that there are enough observations to compare the studied relationship in different countries, which will be useful when analysing when wage flexibility is needed.

The data used for the estimation of the regression includes 11 countries over 11 years. The countries are the initial members of the Eurozone without Luxemburg. The choice of Eurozone countries is not due to the fact that the results have implications for currency areas. The reasons are two: data availability of bilateral trade for long periods and lack of exchange rate fluctuations that affect trade. Given sufficient data, countries with currencies pegged to Euro could be included too. Also, as Tavlas (Tavlas 1993) writes, there is time-inconsistency between the desired characteristics of an economy to be a successful member of a currency area, and the requirements for being in one. His argument is mainly about inflation and inflation is much more susceptible to government policy, but to a lesser extent the same argument can be presented for wage flexibility. In other words: higher wage flexibility can stem from the fact that the country is already in a currency area.

(14)

14

Figure 1 presents export and nominal ULC growth for 2002-2013. Over this period the

correlation between the two variables is very weak, only -0,016.

Figure 1

When we look at Figure 1 and Figure 2, the dynamics of each country a very different picture emerges. Figure 2 0,00% 10,00% 20,00% 30,00% 40,00% 50,00% 60,00% 70,00% 80,00% 90,00% 100,00%

Export and NULC growth 2002-2013

Export growth Nominal ULC growth

-30,00% -20,00% -10,00% 0,00% 10,00% 20,00% 30,00% 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

Export Growth

Austria Belgium Finland France Germany Greece Ireland Italy Netherlands Portugal Spain

(15)

15

Figure 3

Export growth between countries looks very tightly correlated. Immediately obvious is the impact of the crisis in 2009 when there is a strong contraction across the board. In 2010 there is a strong growth almost returning export to pre-crisis levels

Nominal ULC growth shows more variability between countries. It is interesting to observe a spike in ULC growth in 2009 in the hight of the crisis, this is most likely due to the fall of GDP, as GDP is in the numerator of ULC. This is followed by a sharp drop for 2010 to around zero for most countries. Irish ULC drops before all the others in 2009 despite the second biggest fall in GDP for the year at -6,3% and continues to drop in 2010 despite further drop in GDP of over 1%. ULC growth for 2009 and 2010 is -3,6 and -8,8 respectively. This suggests a very strong internal devaluation for the Irish economy. Ireland, and Spain to a lesser degree, are the only countries that exibits such dynamics. Looking at Figure 1 and Figure 2 one can see the problem with using ULC and total export, as discussed when defining ULC, total export is significant part of GDP, which is in the denominator of ULC. This is especially seen during a major shock as the one in 2009-2010 and is evident by the correlations of total export growth and ULC growth, which are very high especially for countries with a high share of export of GDP: Germany -0,915, The Netherlands -0,828, Austria -0,877.

Now turning at the type of data that is used for the estimation of the regressions2. Figure 4 is scatter graf of bilateral export growth and ULC growth differences, shows yet a different story. The trendline is negative as the economic logic predicts, but it has a very small gradient.

2_{Only one regression uses bilateral export growth and ULC growth differences, but they more suitable for}

visualizations. -10 -8 -6 -4 -2 0 2 4 6 8 10 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

Nominal Unit Labour Cost Growth

Austria Belgium Finland France Germany Greece Ireland Italy Netherlands Portugal Spain

(16)

16

Figure 4

3) Results

The results using equation (3) show significant negative relationship between nominal unit labour cost and export as seen in Table 13. The coefficients for the GDP of the exporter and importer are both significant and positive. As FDI has negative values, it is omitted in this regression. The coefficient for credit is not significant. Following from the model, this means that a decline in the unit labour cost of the exporter relative to the importer affects export positively, which is in line with economic logic. The coefficient is however quite small. Also, a positive change in the GDP of the importer has a positive effect on export, as the gravity model suggests.

3_{The time span is 2003-2013.}

-15 -10 -5 0 5 10 15 -0,6 -0,4 -0,2 0 0,2 0,4 0,6 Bil at e ral e xp or t gr owt h

ULC growth diferences

(17)

17

Table 1

In the appendix, equations 2-6, the results are shown for different regressions using (3) ∆𝑈𝐿𝐶 _𝑖𝑗𝑡𝑔𝑟𝑜𝑤𝑡ℎ - the difference in growth rate of ULC instead of ∆𝑈𝐿𝐶_𝑖𝑗𝑡

(4) 𝛽₂(𝑙𝑛𝐺𝐷𝑃_𝑖𝑡+ 𝑙𝑛𝐺𝐷𝑃_𝑗,𝑡) instead of +𝛽₂𝑙𝑛𝐺𝐷𝑃_𝑖𝑡 + 𝛽₃𝑙𝑛𝐺𝐷𝑃_𝑗,𝑡 following (Bussiere et al 2007), this is a way of dealing with the problem with the GDP of the exporter, being a result of the dependant variable.

(5) both ∆𝑈𝐿𝐶 _𝑖𝑗𝑡𝑔𝑟𝑜𝑤𝑡ℎand 𝛽₂(𝑙𝑛𝐺𝐷𝑃_𝑖𝑡+ 𝑙𝑛𝐺𝐷𝑃_𝑗,𝑡) (6) growth rates of all the variables

The coefficient for unit labour cost is negative and significant in all of the regressions. The coefficients for GDP are positive and significant and the coefficients for credit and FDI are not significant.

Replacing export with import in equation (2) changes the coefficient of ∆𝑈𝐿𝐶_𝑖𝑗𝑡form negative to positive while still being significate4. This means that not only does unit labour cost has effect on export, but also on import. The coefficient here as well is quite small.

4_{See table 19 in the Appendix.}

F test that all u_i=0: F(109, 1096) = 371.21 Prob > F = 0.0000 rho .98728192 (fraction of variance due to u_i)

sigma_e .14939127 sigma_u 1.3162397 _cons .1871849 1.3324 0.14 0.888 -2.427158 2.801528 lncrdi -.0095384 .0139925 -0.68 0.496 -.0369936 .0179167 lngdpj 1.340977 .1026164 13.07 0.000 1.13963 1.542323 lngdpi .3203177 .1026328 3.12 0.002 .1189388 .5216967 dulcgrij -.0123094 .0017326 -7.10 0.000 -.0157091 -.0089097 lnexpij Coef. Std. Err. t P>|t| [95% Conf. Interval] corr(u_i, Xb) = -0.0793 Prob > F = 0.0000 F(4,1096) = 110.05 overall = 0.4702 max = 11 between = 0.4718 avg = 11.0 R-sq: within = 0.2865 Obs per group: min = 11 Group variable: id1 Number of groups = 110 Fixed-effects (within) regression Number of obs = 1210

(18)

18

As mentioned, one of the benefits of using bilateral trade data is the opportunity to estimate a regression for each individual country. The results are summarized in Table 2.5

Table 2

The coefficient for ULC is negative and significant for Belgium, Finland, Greece, Italy and Portugal. For the rest it is not significant. The coefficient of the GDP of the exporter is positive and significant for all the countries except Austria, Finland, Ireland and Portugal. The coefficient for the GDP of the importer is positive and significant for all, but Greece, where it is not significant. These results are curious as there is not a clear divide core-periphery or small-big countries. Also their export as a percentage of GDP varies widely from around 30% percent for Italy and Greece to above 80% for Belgium. From a statistical point of view this is a welcome development as it shows that the results are not significant due to the presence of export in GDP and GDP in ULC. Still one can see that the countries where ULC plays a role in exports are either from the periphery like Greece, Italy, and Portugal or smaller like Belgium and Finland.

Even though the results are statistically significant I would caution against taking the results as an indisputable evidence for a strong link between ULC difference and bilateral export. The coefficients are small, the issue of export being in ULC, while small, is still present and there is no apparent explanation why the results are significant for some countries and not for others. The same goes for the coefficient for the GDP of the exporter which is problematic to analyse, as bilateral export is a component of GDP.

The interpretation of the results is that for Belgium, Finland, Greece, Italy and Portugal will be beneficial to lower their cost.

5_{10 groups over 11 years, 110 observations. For details look at tables 8-18 in the Appendix.}

ULC GDPi GDPj

Coef. P>|t| Coef. P>|t| Coef. P>|t| AT 0.0077163 0.102 0.384484 0.636 2.262843 0 BE -0.0086752 0 1.476501 0 1.301794 0 DE 0.0020229 0.386 0.594395 0.011 1.964963 0 ES -0.004942 0.07 0.936513 0.007 1.385491 0 FI -0.0257138 0 -0.3355 0.542 1.842484 0 FR 0.0004743 0.858 -1.03685 0.006 1.372154 0 GR -0.0189577 0 0.873115 0 0.565016 0.268 IE -0.0000665 0.984 -0.29237 0.217 0.988932 0.002 IT -0.0050738 0.05 2.400813 0 1.907698 0 NL -0.0033572 0.096 2.541778 0 1.206776 0 PT -0.0105944 0.021 -0.63617 0.445 0.788778 0.052

(19)

19

VI. Conclusion

So the question posed whether changes in wages have effect on the changes in exports does not have a definite answer. The coefficient is negative and significant for the sample of 11 countries as a whole. The coefficients for the individual countries are negative and significant for Belgium, Finland, Greece, Italy and Portugal and nonsignificant for the rest. The results suggest that in theory these countries can boost their exports by moderating their wages, i.e. their export has some elasticity to the local wages. But how elastic is it? The coefficients in this study are quite small, but for definite results a more detailed regression has to be constructed.

VII. References:

Agell, Jonas and Helge Bennmarker (2003). “Endogenous wage rigidity”, CESifo Working

Paper No. 1081

Alesina, Alberto and Enrico Spolaore and Romain Wacziag (2005). “Trade, Growth and the Size of Countries”, Handbook of Economic Growth, ed. 1, vol.1, num. 1

Arpaia, Alfonso and Karl Pichelmann (2007). “Nominal and real wage flexibility in EMU”, European economy. Economic Papers, No 281

Azariadis, Costas and Joseph E. Stiglitz (1983)."Implicit Contracts and Fixed Price Equilibria," Quart. J. Econ., Supplement, 1983, 98(3), pp. 1-22.

Bertola, G., Dabusinskas, A., Hoeberichts, M., Izquierdo, M., Kwapil, C., Montornès, J. and R. Radowski (2010). "Price, wage and employment response to shocks evidence from the WDN survey", ECB Working paper series No 1164

Babecký, J., Du Caju, P., Kosma, D., Lawless, M., Messina, J. and T. Rõõm (2012). "How do European firms adjust their labour costs when nominal wages are rigid?", Labour Economics, Vol. 19, No 5, p. 792-801

Babecký, J., Du Caju, P., Kosma, D., Lawless, M., Messina, J. and T. Rõõm (2009). "Downward nominal and real wage rigidity: Survey evidence from European firms?",ECB Working paper series No 1105

Baltagi, Bati, Peter Egger and Michael Pfaffermayr (2014).”Panel Data Gravity Models of International Trade” CESifo Working Paper No. 4616

Bergstrand, Jeffrey (1985). “The Gravity Equation In International Trade: Some

(20)

20

and Statistic, Vol.67, No. 3 (Aug., 1985), pp. 474-481

Blanchard, Olivier Jean and Lawrence F. Katz, Robert E. Hall, Barry Eichengreen. (1992). “Regional Evolutions”, Brookings Papers on Economic Activity, Vol. 1992, No. 1, 1-75

Blanchard, Olivier Jean and Lawrence F. Katz (1999). “Wage Dynamics: Reconciling Theory and Evidence”, NBER Working Paper No. 6924

Blanchard, Olivier and Thomas Philippon. (2004). “The Quality of Labor Relations and Unemployment” NBER Working Paper No. 10590, JEL No. E24, J5, J64

Blanchard, Olivier and Jordi Galí. (2005). “Real Wage Rigidities and the New Keynesian Model” NBER Working Paper No. 11806, JEL No. E32, E50

Blanchflower, D., A. Oswald and M. Garrett. (1990). “Insider Power in Wage Determination”, Economica 57. 143-170.

Blanchflower, D. and A. Oswald. (1995) “An Introduction to the Wage Curve”, Journal

Of Economic Perspectives, Summer, 9(3), 153-167.

Branson William H, and Julio J. Rotemberg. (1980) “International adjustment with wage Rigidity”, European Economic Review, 13(3):309–337

Buono, Ines and Sara Formai (2014). “Bank Credit and Firm Export: Is There Really a Link?” Banca d’Italia, June 2014

Bussiere, Matthieu, Jarko Fidrmuc and Bernd Schnatz (2005) “Trade Integration of Central and Eastern European Countries. Lessons from a Gravity Model”, ECB

Working Paper No. 545

Chor, Davin and Kalina Manova (2012) "Off the cliff and back? Credit conditions and international trade during the global financial crisis”, Journal of International

Economics, vol. 87, issue 1, p.117-133

Chrisoffel, Kai and Tobias Linzert (2005) “The role of real wage rigidity and labor

market frictions for unemployment an inflation dynamics”, ECB Working paper eries No 556

Decressin, Jorg and Antonio Fatas. (1995) “Regional labor market dynamics in Europe.

European Economic Review 39, 1627-1655

De Grauwe, Paul and Schnabl, Gunther. (2004). “Nominal versus Real Convergence with Respect to EMU Accession: How to Cope with the Balassa-Samuelson Dillema,

(21)

21

De Grauwe, Paul and Francesco Paolo Mongelli “Endogeneities of Optimum Currency Areas” ECB Working paper series No 468

Dickens et all (2007) “How Wages Change: Micro Evidence from the International Wage Flexibility Project” Federal Reserve Bank of New York Staff Reports, no. 275

Druant M., S. Fabiani, G. Kezdi, A. Lamo, F. Martins and R. Sabbatini, (2009) "How are firms' wages and prices linked: Survey evidence in Europe", ECB Working paper seriesNo 1084

Galuscak K., Keeney, M., Nicolitsas, D., Smets, F., Strzelecki, P. and M.

Vodopivec (2009) "The determination of wages of newly hired employees: survey evidence on internal versus external factors", ECB Working paper series

No 1153

Fehr, Ernst and Götte, Lorenz (2007). “Robustness and Real Consequences of Nominal Wage Rigidity”, Kieler Arbeitspapiere, No. 1343

Felipe, Jesus and Utsav Kumar (2011). “Unit Labor Cost in the Eurozone: The Competitiveness Debate Again”, Levi Economics Institute Working paper No. 651

Holmlund, Bertil and Johnny Zetterberg (1991). “Insider effects in wage determination Evidence from five countries”, European Economic Review 35: 1009-1034

Huizinga, Frederik and Peter Broer (2004). “Wage moderation and labour productivity”,

CPB Discussion Paper, No. 28

Keynes, John Maynard (1936) General theory of Interest, Palgrave Macmillan,

Kniesner, Thomas J.and Arthur H. Goldsmith (1987). “A Survey of Alternative Models

of the Aggregate U.S. Labor Market”, Journal of Economic Literature, Vol. 25, No. 3, 1241-1280

Kutan, Ali M and Goran Vuksic (2007). “Foreign Direct Investment and Export Performance: Empirical Evidence”,Comparative Economic Studies 49, pp. 430-445

Mundell, Robert A. (1961). “A Theory of Optimum Currency Areas.” American

Economic Review 51: 657-65

Summers, L. H., (1988), “Relative wages, efficiency wages, and Keynesian unemployment” American Economic Review 78, 383-388.

(22)

22

Tavlas, George S. (1993). “The ‘New’ Theory of Optimum Currency Areas”, The World

Economy, 663-685.

Teulings, Coen and Joop Hartog (1998). “Corporatism or Competition? Labour Contracts, Institutions and Wage Structures in International Comparison”,

Cambridge University Press

Vries, Vincent de. (2014). “Unit Labor Costs and the Current Account in the Eurozone” UvA Master Thesis

Weitzman, Martin L. “The share economy”. Harvard U. Press, Cambridge, MA, 1984.

Taylor, John B. "Aggregate Dynamics and Staggered Contracts," J. Polit. Econ., Feb. 1980, 88(1), pp. 1- 23.

VIII. Appendix

The countries in the sample are Austria, Belgium, Germany, Finland, France, Greece, Ireland, Italy, the Netherlands, Portugal and Spain. The years are 2003-2013, except for regression (5) 2004-2013, which regresses growth of export.

Tables 8-18 present the results for each separate country.

Table 19 presents the results for import, using equation (2), changing 𝑙𝑛𝐸𝑖𝑗𝑡 with 𝑙𝑛𝐼𝑀𝑃𝑖𝑗𝑡 The data for the bilateral exports, nominal unit labour cost, real GDP and FDI are from Eurostat.

The data about bank credit is compiled from data published by ECB, about the monthly new loans to non-financial corporations.

These are the results for the all 11 countries, each with 10 trading partners – 110 country pairs over 11 years - 1210 observations in total.

(23)

23 2. 𝑙𝑛𝐸𝑖𝑗𝑡 = 𝛼 + 𝛽1∆𝑈𝐿𝐶𝑖𝑗𝑡 +𝛽2𝑙𝑛𝐺𝐷𝑃𝑖𝑡+ 𝛽3𝑙𝑛𝐺𝐷𝑃𝑗,𝑡+ 𝛽4𝑙𝑛𝐶𝑖𝑡+ 𝜀𝑖𝑗𝑡 (2) 3. 𝑙𝑛𝐸𝑖𝑗𝑡 = 𝛼 + 𝛽1∆𝑈𝐿𝐶 𝑖𝑗𝑡𝑔𝑟𝑜𝑤𝑡ℎ +𝛽2𝑙𝑛𝐺𝐷𝑃𝑖𝑡 + 𝛽3𝑙𝑛𝐺𝐷𝑃𝑗,𝑡+ 𝛽4𝑙𝑛𝐶𝑖𝑡+ 𝜀𝑖𝑗𝑡 (3) 4. 𝑙𝑛𝐸𝑖𝑗𝑡 = 𝛼 + 𝛽₁∆𝑈𝐿𝐶_𝑖𝑗𝑡 +𝛽₂(𝑙𝑛𝐺𝐷𝑃_𝑖𝑡 + 𝑙𝑛𝐺𝐷𝑃_𝑗,𝑡) + 𝛽₄𝑙𝑛𝐶_𝑖𝑡+ 𝜀𝑖𝑗𝑡 (4)

Here I am following (Bussiere et al 2007), where they sum the logs of the GDP F test that all u_i=0: F(109, 1096) = 371.55 Prob > F = 0.0000 rho .98860636 (fraction of variance due to u_i)

sigma_e .14890926 sigma_u 1.3870824 _cons .4242899 1.329162 0.32 0.750 -2.1837 3.032279 lncrdi .0079246 .0144553 0.55 0.584 -.0204386 .0362878 lngdpj 1.409156 .0976039 14.44 0.000 1.217644 1.600668 lngdpi .2184113 .0970428 2.25 0.025 .0280007 .408822 dulcinij -.0069371 .0009116 -7.61 0.000 -.0087258 -.0051484 lnexpij Coef. Std. Err. t P>|t| [95% Conf. Interval] corr(u_i, Xb) = -0.1630 Prob > F = 0.0000 F(4,1096) = 112.54 overall = 0.4242 max = 11 between = 0.4254 avg = 11.0 R-sq: within = 0.2911 Obs per group: min = 11 Group variable: id1 Number of groups = 110 Fixed-effects (within) regression Number of obs = 1210

sigma_e .14939127 sigma_u 1.3162397 _cons .1871849 1.3324 0.14 0.888 -2.427158 2.801528 lncrdi -.0095384 .0139925 -0.68 0.496 -.0369936 .0179167 lngdpj 1.340977 .1026164 13.07 0.000 1.13963 1.542323 lngdpi .3203177 .1026328 3.12 0.002 .1189388 .5216967 dulcgrij -.0123094 .0017326 -7.10 0.000 -.0157091 -.0089097 lnexpij Coef. Std. Err. t P>|t| [95% Conf. Interval] corr(u_i, Xb) = -0.0793 Prob > F = 0.0000 F(4,1096) = 110.05 overall = 0.4702 max = 11 between = 0.4718 avg = 11.0 R-sq: within = 0.2865 Obs per group: min = 11 Group variable: id1 Number of groups = 110 Fixed-effects (within) regression Number of obs = 1210

(24)

24

5. 𝑙𝑛𝐸𝑖𝑗𝑡 = 𝛼 + 𝛽₁∆𝑈𝐿𝐶 _𝑖𝑗𝑡𝑔𝑟𝑜𝑤𝑡ℎ +𝛽₂(𝑙𝑛𝐺𝐷𝑃_𝑖𝑡+ 𝑙𝑛𝐺𝐷𝑃_𝑗,𝑡) + 𝛽₄𝑙𝑛𝐶_𝑖𝑡+ 𝜀𝑖𝑗𝑡

(5)

6. 𝐸 _𝑖𝑗𝑡𝑔𝑟𝑜𝑤𝑡ℎ = 𝛼 + 𝛽₁∆𝑈𝐿𝐶 _𝑖𝑗𝑡𝑔𝑟𝑜𝑤𝑡ℎ +𝛽₂𝐺𝐷𝑃 _𝑖𝑡𝑔𝑟𝑜𝑤𝑡ℎ + +𝛽₃𝐺𝐷𝑃 _𝑗𝑡𝑔𝑟𝑜𝑤𝑡ℎ + 𝛽₄𝐹𝐷𝐼 _𝑖𝑡𝑔𝑟𝑜𝑤𝑡ℎ +

𝐶𝑅𝐷 _𝑖𝑡𝑔𝑟𝑜𝑤𝑡ℎ + 𝜀𝑖𝑗𝑡 (6)

Here I regress growth of all the variables over 10 years, so 1100 observations. F test that all u_i=0: F(109, 1097) = 389.85 Prob > F = 0.0000 rho .97931094 (fraction of variance due to u_i)

sigma_e .1523559 sigma_u 1.0482123 _cons .455985 1.359918 0.34 0.737 -2.21235 3.12432 lncrdi .0105894 .0147851 0.72 0.474 -.0184209 .0395996 lngdpilngdpj .8113818 .0531963 15.25 0.000 .7070038 .9157598 dulcinij -.0092793 .000872 -10.64 0.000 -.0109902 -.0075684 lnexpij Coef. Std. Err. t P>|t| [95% Conf. Interval] corr(u_i, Xb) = 0.4167 Prob > F = 0.0000 F(3,1097) = 126.66 overall = 0.7174 max = 11 between = 0.7218 avg = 11.0 R-sq: within = 0.2573 Obs per group: min = 11 Group variable: id1 Number of groups = 110 Fixed-effects (within) regression Number of obs = 1210

sigma_e .1515735 sigma_u 1.049061 _cons .1863883 1.351862 0.14 0.890 -2.466139 2.838916 lncrdi -.0095921 .0141969 -0.68 0.499 -.0374483 .018264 lngdpilngdpj .8307016 .0527881 15.74 0.000 .7271246 .9342785 dulcgrij -.0171975 .0015334 -11.22 0.000 -.0202063 -.0141887 lnexpij Coef. Std. Err. t P>|t| [95% Conf. Interval] corr(u_i, Xb) = 0.4010 Prob > F = 0.0000 F(3,1097) = 131.76 overall = 0.7127 max = 11 between = 0.7170 avg = 11.0 R-sq: within = 0.2649 Obs per group: min = 11 Group variable: id1 Number of groups = 110 Fixed-effects (within) regression Number of obs = 1210

(25)

25

The following regressions are for each individual country, using equation (2)

𝑙𝑛𝐸𝑖𝑗𝑡 = 𝛼 + 𝛽₁∆𝑈𝐿𝐶_𝑖𝑗𝑡 +𝛽₂𝑙𝑛𝐺𝐷𝑃_𝑖𝑡+ 𝛽₃𝑙𝑛𝐺𝐷𝑃_𝑗,𝑡+ 𝛽₄𝑙𝑛𝐶_𝑖𝑡 + 𝜀𝑖𝑗𝑡

8. AT

sigma_e .12661674 sigma_u .03495162 _cons -.0078896 .0040072 -1.97 0.049 -.0157531 -.000026 crdi -.0205037 .0177018 -1.16 0.247 -.0552413 .014234 fdii .0000477 .0002363 0.20 0.840 -.0004161 .0005115 gdpj 2.075057 .1797262 11.55 0.000 1.722367 2.427748 gdpi .9755308 .1841984 5.30 0.000 .6140645 1.336997 dulcgrij -.0037473 .0013382 -2.80 0.005 -.0063732 -.0011213 expij Coef. Std. Err. t P>|t| [95% Conf. Interval] corr(u_i, Xb) = -0.0748 Prob > F = 0.0000 F(5,985) = 90.16 overall = 0.2986 max = 10 between = 0.1153 avg = 10.0 R-sq: within = 0.3140 Obs per group: min = 10 Group variable: id1 Number of groups = 110 Fixed-effects (within) regression Number of obs = 1100

sigma_e .18060832 sigma_u 1.120821 _cons -10.22014 8.294957 -1.23 0.221 -26.6855 6.245219 lncrdi -.2793718 .1660141 -1.68 0.096 -.6089072 .0501636 lngdpj 2.262843 .448183 5.05 0.000 1.373207 3.152479 lngdpi .3844837 .8090114 0.48 0.636 -1.221391 1.990359 dulcinij .0077163 .0046708 1.65 0.102 -.001555 .0169877 lnexpij Coef. Std. Err. t P>|t| [95% Conf. Interval] corr(u_i, Xb) = -0.9034 Prob > F = 0.0000 F(4,96) = 7.94 overall = 0.8869 max = 11 between = 0.8995 avg = 11.0 R-sq: within = 0.2485 Obs per group: min = 11 Group variable: id1 Number of groups = 10 Fixed-effects (within) regression Number of obs = 110

(26)

26 9. BE

10. DE

sigma_e .09784238 sigma_u .58679692 _cons -9.967859 3.018033 -3.30 0.001 -15.95861 -3.977111 lncrdi -.2629157 .1151159 -2.28 0.025 -.491419 -.0344124 lngdpj 1.301794 .240108 5.42 0.000 .8251834 1.778405 lngdpi 1.476501 .3289668 4.49 0.000 .8235072 2.129495 dulcinij -.0086752 .0023727 -3.66 0.000 -.013385 -.0039654 lnexpij Coef. Std. Err. t P>|t| [95% Conf. Interval] corr(u_i, Xb) = -0.1049 Prob > F = 0.0000 F(4,96) = 33.58 overall = 0.8422 max = 11 between = 0.8448 avg = 11.0 R-sq: within = 0.5832 Obs per group: min = 11 Group variable: id1 Number of groups = 10 Fixed-effects (within) regression Number of obs = 110

sigma_e .08613031 sigma_u 1.1360464 _cons -10.67753 4.102929 -2.60 0.011 -18.82178 -2.533282 lncrdi .0176701 .0586378 0.30 0.764 -.0987249 .1340652 lngdpj 1.964963 .2192305 8.96 0.000 1.529794 2.400132 lngdpi .5943949 .2301742 2.58 0.011 .1375028 1.051287 dulcinij .0020229 .0023219 0.87 0.386 -.0025859 .0066317 lnexpij Coef. Std. Err. t P>|t| [95% Conf. Interval] corr(u_i, Xb) = -0.8214 Prob > F = 0.0000 F(4,96) = 44.57 overall = 0.6684 max = 11 between = 0.6714 avg = 11.0 R-sq: within = 0.6500 Obs per group: min = 11 Group variable: id1 Number of groups = 10 Fixed-effects (within) regression Number of obs = 110

(27)

27 11. ES

12. FI

sigma_e .0984471 sigma_u .99324494 _cons -6.5186 3.456618 -1.89 0.062 -13.37993 .342732 lncrdi -.1695362 .0413459 -4.10 0.000 -.2516071 -.0874653 lngdpj 1.385491 .2514462 5.51 0.000 .8863738 1.884607 lngdpi .9365125 .3415678 2.74 0.007 .2585057 1.614519 dulcinij -.004942 .0026951 -1.83 0.070 -.0102917 .0004076 lnexpij Coef. Std. Err. t P>|t| [95% Conf. Interval] corr(u_i, Xb) = -0.4044 Prob > F = 0.0000 F(4,96) = 32.16 overall = 0.5856 max = 11 between = 0.5862 avg = 11.0 R-sq: within = 0.5727 Obs per group: min = 11 Group variable: id1 Number of groups = 10 Fixed-effects (within) regression Number of obs = 110

sigma_e .17580473 sigma_u 1.0659006 _cons 1.097617 5.848916 0.19 0.852 -10.51239 12.70762 lncrdi -.0687215 .0960355 -0.72 0.476 -.2593504 .1219074 lngdpj 1.842484 .4422655 4.17 0.000 .9645941 2.720374 lngdpi -.3354996 .5477651 -0.61 0.542 -1.422805 .7518057 dulcinij -.0257138 .0028184 -9.12 0.000 -.0313084 -.0201193 lnexpij Coef. Std. Err. t P>|t| [95% Conf. Interval] corr(u_i, Xb) = -0.8217 Prob > F = 0.0000 F(4,96) = 34.10 overall = 0.7398 max = 11 between = 0.7534 avg = 11.0 R-sq: within = 0.5869 Obs per group: min = 11 Group variable: id1 Number of groups = 10 Fixed-effects (within) regression Number of obs = 110

Wage flexibility and export in a currency area

Wage Flexibility and Export in a Currency Area

Contents

I.

Introduction.

II.

The theory of optimum currency areas.