Impact of euro-area membership on structural reforms in product market and business regulation

: This paper investigates whether membership in the euro area enhances or hampers structural reforms in the regulation of product markets and business environment. It updates the empirical results of the early literature and adds to it by applying new indicators for structural reforms. By estimating a dynamic panel data model with system GMM, I find that euro-area countries deregulate significantly more than other OECD countries in product markets. This result is confirmed applying the longer panel for network sector regulation. However, I do not find any evidence for reforms in business regulation that would ease doing business in a cross-section analysis.


Introduction
Prior to the creation of the euro area, there was a lively debate among scholars whether the euro would hamper or speed up structural reforms in the regulation of product and labour markets. In that debate, the TINA (There Is No Alternative) argument was predominant.
However, the conclusions of theoretical models concerning this question are rather ambiguous or contradictory. Hence, right after the introduction of the euro, several studies were conducted to test the hypotheses empirically.
In my view, it is time to reassess the empirical relationship between the introduction of the euro and structural reforms with the substantial larger sample period now available, more than a decade after the creation of the euro area. Moreover, I apply indicators that were not available or usable for the early literature. In this paper, I do not assess the impact of the euro on labour markets. Rather, I focus on the effect on product market and business regulation. We will look at whether euro-area membership reduces product market regulation and improves the ease of doing business.
To begin with, I use the OECD's economy-wide product market regulation (PMR) indicator to estimate a dynamic panel data model with system GMM, where a dummy variable for the euro-area membership of a country is the variable of interest. Then, I use a difference-indifferences framework to estimate the euro's impact on network sector regulation in a long panel of OECD countries-with indicators already applied by the early literature on structural reforms. This enables me to address the issues of identification and country fixed effects.
Finally, I perform a simple OLS cross-sectional analysis with the World Bank's Doing Business distance to frontier indicator. To summarize the results, I find a significant and robust effect of euro-area membership on the deregulation of product markets and, more specific, of network sectors. However, the analysis does not permit to say that the introduction of the euro facilitates or complicates doing business.
This paper contributes to the early literature on structural reforms. Alesina, Ardagna and Galasso (2010) investigate whether or not the introduction of the euro has accelerated structural reforms. They call the deregulation in product markets, and the liberalization and deregulation in the labour markets structural reforms. They disentangle the effect of the euro and the European Single Market, and find significant correlations between the speed of deregulation in product markets and the adoption of the euro. Moreover, the effect of the euro on product market regulation was larger the larger the initial level of regulation. The results are less clear with respect to the labour markets. They find evidence for wage moderation prior the adoption of the euro, and mixed results thereafter. Due to data availability at the time, their results concerning the deregulation in product markets are predominantly based upon indicators on regulation of seven network industries (electricity and gas supply, road freight, air passenger transport, rail transport, post, and telecommunications). This set of indicators is to some extent a forerunner of the OECD Product Market Regulation (PMR) indicator that I use in my first regression analysis (Section 3). The main difference between these sets of indicators is the area of regulation: the first measure regulation in specific sectors, the latter measures economy-wide regulation.
Their second limitation is the observable period after the introduction of the euro. The latest data on regulation they use is from 2003. Duval and Elmeskov (2005) conduct a similar empirical analysis, where they explore the role of the monetary regime for structural reforms. They estimate the likelihood that countries undertake reforms in either product market regulation or in one of four specific areas of labour market policies, depending on whether the country is engaged in a fixed exchange rate arrangement or not. The dataset is basically the same as in Alesina, Ardagna and Galasso (2010): the OECD dataset on labour market reform and the OECD indicators of product market regulation in network sectors for 21 OECD countries for the period [1985][1986][1987][1988][1989][1990][1991][1992][1993][1994][1995][1996][1997][1998][1999][2000][2001][2002][2003]. Based on a panel probit regression, there is weak evidence that countries with a fixed exchange rate regime undertake less structural reforms. Duval and Elmeskov (2005) restrict their analysis to major reforms as opposed to small ones. By this, they empirically prove the theoretical conclusion of Saint-Paul and Bentolila (2001) (see Section 2) that the participation in the euro area makes big reforms more unlikely. However, Duval's and Elmeskov's (2005) analysis cannot identify the effect of the euro area on rather small, but gradual reforms, which should be encouraged (Saint-Paul and Bentolila 2001). Belke, Herz and Vogel (2007) and Belke and Vogel (2015) also contribute to this strand of literature by focusing on monetary commitment rather than on the introduction of the euro. In a sample of 23 OECD countries from 1970 to 2000, Belke, Herz and Vogel (2007) find evidence in favour of the TINA argument for reforms in labour and product markets. Belke and Vogel (2015) gain mixed evidence on the relationship between market-oriented structural reforms as economic liberalization and monetary commitment in transition countries.
In a recent paper, Schönfelder and Wagner (2016)  The paper is organized as follows. Section 2 discusses how euro-area membership could influence structural reforms. The empirical analyses of the three regulation indicators are divided in separate sections-each explaining the data, the estimation approach and the results. Section 3 studies the impact of euro-area membership on economy-wide product market regulation. Network sector regulation is evaluated in Section 4. In Section 5, I assess the euro's impact of the ease of doing business. Section 6 concludes.

2
Theoretical arguments of why euro-area membership could influence structural reforms Alesina, Ardagna and Galasso (2010, p. 2) lay down some economic arguments why the membership in the euro area could accelerate and facilitate deregulation and liberalization in product and labour markets. In this discussion, I focus on their arguments concerning product market deregulation. They describe two channels by which the euro could foster reforms: the competition channel and the adjustment channel. The former establishes a relationship between more competition due to the single market and the cost of regulation in product markets. A single currency likely increases price transparency, and may expose the cost of structural rigidities more obviously. Combining greater price transparency with more competition within the European single market could make it more difficult for domestic monopolist to protect their rents. The resulting difficulties of local monopolist to dominate regional markets could create pressure for deregulation of product markets.
One recent example of widespread domestic regulatory protection is the German legislation restricting internet sales of medical products for human use by pharmacies. The second channel becomes relevant whenever a country is losing competitiveness. The common currency eliminates the possibility of strategic devaluations. Domestic firms that are producing tradeable goods and their special interest groups could demand deregulation of the markets for non-tradeable input goods, e.g. services and transportation, to contain costs.
Hence, the tradeable sector reacts directly to competition and translates this pressure to the intermediate goods producers (Alesina, Ardagna and Galasso 2010, pp. 2-7;Duval and Elmeskov 2005, p. 10). This argument is related to the TINA (There Is No Alternative) argument. By introducing the euro, the member countries become unable to use monetary policy to accommodate asymmetric shocks. Instead, adjustment has to come via a boom or recession. Therefore, euro-area members have to develop market-based adjustment channels to adjust to shocks through changes in prices (Alesina, Ardagna and Galasso 2010, p. 6;Bean 1998, p. 368).
In a case study, Fernández-Villaverde, Garicano and Santos (2013) show that economic reforms were abandoned and institutions deteriorated after the introduction of the euro in Spain, Ireland, Greece, and Portugal. They argue that as the euro facilitated large inflows of capital, which enabled the emergence of the financial bubble in peripheral countries, economic reforms were abandoned, institutions deteriorated, and the response to the credit bubble was delayed. This hampered the growth prospects of these countries. Fernández-Villaverde, Garicano and Santos (2013) analyse two channels to explain this development.
They appear in stark contrast with the German case. First, capital flows relaxed the economic constraints under which agents (e.g., a government or bank manager) were acting, which reduced the pressure for reforms. Second, capital inflows hindered the principal (e.g., voters, shareholders, investors) in extracting signals about the performance of the agent.
Germany did not experience a loosening of its financing conditions because of the introduction of the euro, and it was faced with a stagnant economy. Hence, Germany implemented far-reaching structural reforms, so that the divergence in economic policies and institutions between Germany and the other peripheral countries increased after the introduction of the euro (Fernández-Villaverde, Garicano and Santos 2013, pp. 146-147).
Other factors than the euro may influence reforms on regulation in product markets and business environment. As far as these factors are correlated with the membership in the euro area, they have to be included in the regression to avoid the omitted variable bias. Moreover, they are interesting on their own right. Alesina, Ardagna and Galasso (2010) and Duval and Elmeskov (2005) summarize some factors that could create incentives for governments to adopt regulatory reforms. Firstly, according to the TINA argument, regulatory reforms are required to regain international competitiveness when the devaluation channel is not available. Secondly, countries that devaluated more often in the past could be more in need of regulatory reforms. Thirdly, it has been frequently observed that governments implement reforms in response to a macroeconomic or fiscal crisis. This is because governments can overcome a "status quo" bias in crisis situations. In normal times the "status quo" bias tends to prevent growth-enhancing deregulation from being implemented. On the other hand, a sound fiscal position enables governments to compensate losers from regulatory reforms, hence facilitating reforms. Fourthly, high unemployment could facilitate deregulation because there are fewer employees that oppose to bear the short-run costs of deregulation.
Moreover, when unemployment benefits are high, the short-run costs of deregulation are lower for those that could be dismissed. Fifthly, there could be some effect of employment protection legislation on regulation in product markets and business environment, as implementing reforms in one field may pave the way for reforms in others. Finally, the government party-orientation could play a role in the propensity to deregulate.

3
Economy-wide product market regulation

The data set and some statistics
The OECD product market regulation database comprises indicators for economy-wide regulation, sector regulation, regulatory impact, internet regulation, sector regulators (regulatory management practice), and competition law and policy. The indicator for economy-wide regulation (Koske et al. 2015) is particularly interesting as it summarizes a wide array of different regulatory provisions across countries in areas of the product markets where competition is viable. It measures the degree to which policies promote or inhibit competition. The literature has shown that intensive competition in product markets spurs economic growth through increasing productivity and employment. 3 The economy-wide product market regulation (PMR) indicator also integrates data on sectoral regulation (as regards the energy, transportation, communication, retail distribution, and professional services sector) into a comprehensive measure of product market regulation to assess the overall regulatory stance (Wölfl et al. 2009, p. 9). The data on sectoral regulation for networks (energy, transport and communications regulation -ETCR) will be used in a subsequent analysis as a robustness check.  Ochel and Röhn (2006, pp. 49-50). Focusing on institutional environment for business, the World Bank conducts the Business Environment and Enterprise Performance Survey every four years. It provides firm-level data on issues like business-government relations, firm financing, and so on. All of the above data sets rely on surveys or expert interviews for indices on institutional aspects. 5 For a detailed description of the PMR indicator, see Koske et al. (2015). 6 These are Australia, Austria, Belgium, Canada, Chile, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Israel, Italy, Japan, Korea, Luxembourg, Mexico, Netherlands, New Zealand, Norway, Portugal, Spain,  Table 2 presents the changes of the indicator between 1998 and 2013 depending on whether a country participates in the euro area. Euro-area members experienced a pronounced reduction of policies that inhibit competition. The reduction is about twice as large as for the control group (with respect to the median, the mean, and the third quartile). Three-quarters of the euro-area countries improved regulation by at least 0.9 units on a scale from zero to six.     (Beck et al. 2001), and is made available by the Quality of Government Institute (Teorell et al. 2013). The party orientation is coded along the following criteria: (1) right for parties that are defined as conservative, Christian democratic, or rightwing; (2) center for parties that are defined as centrist or when party position can best be described as centrist (e.g., party advocates strengthening private enterprise in a social-liberal context); and (3) left for parties that are defined as communist, socialist, social democratic, or left-wing. For the use in the PMR analysis, I assign the party orientation of the largest government party to the five-year period, which was in government most of the time (three of the five years). 8

The econometric model and its estimation
In this section, I use a panel model to estimate the impact of euro-area membership on economy-wide product market regulation. The empirical approach is to compare the end-ofperiod levels of the indicator with the levels at the beginning of each period. The differences between both are the period changes. If the period changes are significantly different for euro-area countries in comparison to other OECD countries (when controlling for other determinants), this is evidence for the influence of the euro on regulation. For the PMR indicator, one period corresponds to five years. The estimation equation is To select the appropriate estimator for Equation (1), one usually first discusses whether fixed effects are present or the individual effects are uncorrelated to the regressors (random effects). However, as the variable of interest, euro, does not change its value over the sample period (time-invariant), it is impossible to estimate the coefficient of interest with a fixed effect estimator. In the following, we assume a random effects model, although one has to bear in mind that this assumption is restrictive.
If one regressor is correlated with the error term, the least squares estimates of the coefficients are inconsistent. This is called the simultaneous equation bias (see, e.g., Baltagi 2008, pp. 121-129). By economic reasoning, one can see that the strong exogeneity assumption is not appropriate for every variable. The discussion starts with the variables, for which the strong exogeneity assumption is appropriate and ends with the variables, for which I assume endogeneity. The decision for joining the euro area is not correlated with the error terms, even with the first period errors, because it has been decided upon much earlier.
Therefore, I treat the euro variable as strict exogenous. Strict exogeneity is also appropriate for the variable n.deval, as the devaluations occurred at least several years before the sample start. It is conceivable that the government's election is influenced by past product 9 One can re-write the Equation (1) as market regulatory reforms, which means that the gov variable could be predetermined.
However, I think that the influence of product market regulatory reforms on elections is marginal, which justifies treating the gov variable as strict exogenous as well. With respect to the labour market regulation, it is reasonable to assume strict exogeneity as the gross replacement rate and labour market legislation protection are the expression of social preferences and highly persistent. On the contrary, employment could indeed be influenced by reforms in product markets. Therefore, I assume that it is endogenous. For the remaining control variables, the analysis is complicated not only by endogeneity, but also by opposing signs of the contemporary effect and the effect of the control variables from previous periods.
In Section 2, it was argued that in times of crisis, when the GDP growth is below its potential and fiscal deficits are alarming, governments can more easily change a "status quo" and implement growth-enhancing reforms. This would correspond to a positive sign of the coefficients of fiscal and GAP, and to a negative sign of the coefficient of crisis05 in the PMR analysis. However, within the same, or even the next period, the fiscal and macroeconomic stance could be negatively influenced by increasing product market regulation. This way of argumentation is even more urgent for the competitiveness indicators CPIDR and ULCDR.
On the one hand, governments could seek to implement deregulatory reforms when the loss of international competitiveness is high. On the other hand, there is probably a direct effect of product market regulation on international competitiveness. Deregulation in product markets should lead to lower consumer prices and hence higher international competitiveness than without the deregulation. Therefore, I am seriously concerned about the endogeneity of these two regressors and the identification of two opposing effects.
The inclusion of a lagged depended variable in the estimation equation leads to similar problems as the qualification of the strong exogeneity assumption of regressors. The individual specific effect is correlated with the lagged dependent variable and hence the is the period change.
lagged dependent variable is correlated with the composite error term. Therefore, the least squares estimates of a dynamic panel data model are inconsistent in short panels. This is the so-called Nickell bias (Nickell 1981).
In the panel data analysis of this chapter, the sample period is short (four periods). To deal with potential endogeneity of the regressors, system GMM as proposed by Blundell and Bond (1998) is employed. 10 This estimator consistently estimates dynamic panel data models in short panels. All potentially endogenous variables are instrumented by their second and third lags. Additionally, lagged differences are used as instruments for the equations in levels. However, the number of instruments is high, relative to the sample size. Hence, the finite sample at hand lacks adequate information to estimate the variance matrix of moments well. By this, the first-step and the second-step matrix become singular, which forces the use of the generalized inverse. As Roodman (2009b, p. 98) notes, this does not compromise consistency of the estimated coefficients, but does exaggerates the distance of the feasible efficient GMM estimator from the asymptotical ideal. The number of instruments (including the time dummies) is reported for every regression in Table 3. As a rule of thumb, the number of instruments should not exceed the number of countries, and the p-value of the Hansen-Sargan test should not be close to unity. Both would be symptoms of instrument proliferation (Roodman 2009a).

Econometric evidence
In this section, I estimate the impact of euro-area membership on product market regulation by two-step system GMM (Blundell and Bond 1998) in a dynamic panel-data model.
Equation (1) is estimated with changing sets of control variables. 11 The first regression 10 All the results of this paper are obtained using R version 3.1.1 with the packages plm 1.4-0, lmtest 0.9-33, car 2.0-21 and sandwich 2.3-2 (R Core Team 2014; Croissant and Millo 2008;Zeileis and Hothorn 2002;Fox and Weisberg 2011;Zeileis 2004). 11 See Table 10 for an overview of the expected coefficient's signs.
includes the lagged dependent variable, the euro variable, and control variables that appeared to be significant throughout several specifications. In all specifications, the coefficient of the lagged dependent variable is highly significant and between 0.44 and 0.81.
The coefficient of the variable euro is significantly different from zero in every specification (only at the 10 percent level in the Model 4 and 6). The coefficient has a negative sign and ranges between −0.08 and −0.15 dependent on the specification. Hence, the euro-area countries deregulated significantly more than other countries. In the first regression, the coefficient is −0.10. Hence, the short run effect of euro is −0.10 ( 2  ) and the long-run effect amounts to −0.20 ( 2   1 (1 ) Interestingly, only two control variables are (weakly) significant throughout several specifications: the gross replacement rate (GRR) and the employment protection legislation (EPL). An increase in the gross replacement rate by 10 percentage points is associated with a decrease in the PMR indicator by 0.032 (Model 1) in the short-run. That supports the hypothesis that high unemployment benefits reduce the cost of deregulation for those that could be dismissed and therefore facilitate the implementation of deregulation. The employment protection legislation affects regulatory reforms the other way around. EPL is positively associated with regulation that inhibits competition. Hence, there is some evidence that implementing reforms in labour markets may pave the way for reforms in product markets. From the third regression, I subsequently include additional control variables related to different areas. Loosely speaking, these are policy variables, labour market variables, competitiveness variables, and variables signalling a "crisis". All of them turn out to be insignificant. 12 I find no evidence that product market deregulation is pursued to regain international competitiveness or that governments tend to implement reforms when a 12 To prove that the results of system GMM are robust with respect to the number of instruments, I re-estimated Table 3 with collapsed instruments. Methods for cutting down the number of instruments in panel GMM have been proposed among others by Breitung (1994), Judson and Owen (1999), and Roodman (2009a). However, they come to the expense of efficiency. The only noteworthy difference to the previous system GMM results is that the gross replacement rate (GRR) and employment protection legislation (EPL) are significant with the expected signs only in Model 2. The table is available from the author upon request. macroeconomic or fiscal crisis occurs. I also include an interaction term between the lagged dependent variable and the euro dummy, and test on significance (not reported here).
However, the interaction term is not significant. Hence, regulatory reforms do not depend in a different way on the level of regulation in euro-area countries than in other countries.   (2005) finite-sample correction for standard errors (in parentheses) is applied. The second and third lag of all potentially endogenous variables (PMR, fiscal, unempl, GAP, crisis05, CPIDR, ULCDR) are used as instruments. Included exogenous variables (incl. time dummies) are counted in the total number of instruments as well. (iv) The Arellano-Bond test for AR(2) in first differences cannot be performed because there are estimates of the standard error only for two periods. This is because including a lagged dependent variable and taking first-differences depletes the number of periods. Two periods are not sufficient to perform an AR(2) test.  Table 4. The OECD countries are on the path to abandon policies that inhibit competition in network sectors. The mean of the ETCR indicator is considerably lower in 2013 than in 1999, and substantially lower than in 1975.
Network sectors have been heavily deregulated during the last four decades.  to the control group after the introduction of the euro but not before.  Figure 3 displays boxplots of the ETCR changes before and after the introduction of the euro.
They plot the distributions for euro-area members against the control group. The figure confirms the results of the descriptive analysis. One can see a remarkable difference for the change in network sector regulation in euro-area countries in comparison with the control group, but not before the introduction of the euro. Hence, euro-area countries seem to deregulate more heavily than other OECD countries. is that a government rather conducts painful reforms right after its appointments than before an election. By this, the short-term costs of reforms may die away before the next election.
The additional control variables ESM and elec were both found to promote deregulation by Alesina, Ardagna and Galasso (2010).

The econometric model and its estimation
In the second empirical analysis, I use a difference-in-differences framework to estimate the euro's impact on network sector regulation in a long panel of OECD countries. By this, I can address the issues of identification and country fixed effects. The empirical approach of this chapter is basically the same as in the last section: to compare the end-of-period levels of the indicator with the levels at the beginning of each period. If the period changes are significantly different for euro-area countries in comparison to other OECD countries, this is evidence for the euro's influence on network sector regulation. The estimation equation is i t X is a vector of time-variant control variables, and 3  is the corresponding vector of coefficients. Because of the sample's long time dimension, the estimator of the previous section should not be applied in this section. The generalized method of moments (GMM) procedures, such as the Arellano-Bond estimator, the Arellano-Bover estimator, and the Blundell-Bond estimator are suited to short panels with T fixed and N   (Cameron and Trivedi 2007, p. 744). However, they are not appropriate for long panels that comprise many periods with relatively few individuals. The sample of this section comprises 29 years and 27 countries at the most. Therefore, inference can be based on the assumption that T   . The fixedeffects estimator is generally biased in dynamic models (Nickell 1981). However, as T increases, the fixed-effects estimator becomes consistent (Baltagi 2008, p. 147). 15 Our sample period is sufficiently large so that the bias should be small in this estimation. 16 The long time-dimension enables us to address the issues of identification and country fixed effects. However, coefficients of time-invariant variables cannot be estimated (as the number of devaluations during EMS from 1979 to 1993).

Econometric evidence
In this section, I estimate the impact of euro-area membership on regulation in network sectors (energy, transport and communications) in a dynamic panel-data model. Equation (2) is estimated by the two-way fixed-effects OLS estimator in different specifications. The first three models include inter alia the gross replacement rate (GRR) in first lags. However, this greatly reduces the sample size because there is GRR data only until 2005. The Models 4 to 8 include the gross replacement rate but with its eighth lag to boost the sample without completely skipping the gross replacement rate. The Models 7 and 8 are specifications without the gross replacement rate due to the above-mentioned reasons.
Throughout all specifications, the lagged dependent variable, the euro variable, the crisis dummy and the labour market regulation variables EPL and GRR 17 are statistically significant. Hence, the euro-area countries deregulated significantly more than other countries. Moreover, the negative sign of the coefficient of crisis05 supports the argument that in times of crisis, when the GDP growth is well below its potential, it easier for governments to overcome a "status quo" and implement growth-enhancing reforms. The 15 As the Nickell bias, the bias of weakly exogenous or predetermined regressors is also inversely related to the size of the time dimension Breitung (2015, p. 455). Bias size decreases as the time dimension increases. 16 Alternative estimators have been developed that are asymptotically efficient as T tends to infinity. These are the bias-adjustment and maximum likelihood type estimators. Corrected within-group estimators perform best in dynamic panel data models with moderate to large T. Maximum likelihood estimators may be superior if T is small and the autoregressive coefficient is close to unity. However, the attractive features of bias-adjustment and maximum likelihood-type estimators come at the expense of more restrictive model assumptions Breitung (2015). Most relevant in our context is the restrictive assumption of strictly exogenous regressors. Hence, there is not much to gain in applying these alternative estimators with respect to our coefficients of interest. Taking into account the limited gain in applying these estimators, we favour the approach of estimating Equation (2) by twoway within OLS. employment protection legislation EPL is positively associated with regulation that inhibits competition. GRR affects regulatory reforms the other way around. That supports the hypothesis that high unemployment benefits reduce the cost of deregulation for those that could be dismissed and therefore facilitate the implementation of deregulation. The significance of other control variables is either not robust (left-wing government, competitiveness indicator based on relative unit labour costs, parliamentary or presidential elections, unemployment rate, the European Single Market membership, and the government primary balance) or the control variable is not significant at all (output gap, centrist government). However, besides the unemployment rate, the coefficients of all the above-mentioned controls have the expected sign. To summarize the results, I find a significant and robust effect of euro-area membership on the deregulation in network sectors.  As the PMR indicator, the Doing Business indicators use the readings of laws and regulation in each country to measure the strength of legal institutions relevant to business regulation.
About three-quarters of the data are of this type. To capture the complexity and cost of regulatory process, the Doing Business team collaborates with local experts to estimate time needed for a regulatory process (e.g., starting a business). Cost estimates are usually taken from official fee schedules (World Bank 2013, pp. 21-22).
The Doing Business project has two measures to compare the business regulatory efficiency: (1) the ranking of the "ease of doing business" and (2) the "distance to the frontier" (DTF) measure. The first is a relative ranking of countries according to their regulatory efficiency. The second shows the absolute distance to the "frontier", which is the best performance a country has ever had on each of the Doing Business indicators since the year in which the indicator was first collected. The distance to frontier is scaled from zero to onehundred, where one-hundred represents the frontier. Hence, higher scores indicate more efficient business environment and stronger legal institutions (World Bank 2013, p. 2). The distance to frontier measure is appropriate to cross-section and time-series analysis as it is an absolute measure. The data is back-calculated to adjust for changes in methodology (including the emergence of a new frontier) and revisions in data due to corrections (World Bank 2013, pp. 28-29). 18,19 As acknowledged by Acemoglu et al. (2013, p. 8)      They plot the distributions for euro-area members against the control group. In contrast to the product market regulation, there is almost no difference in the change in business regulation in euro-area countries compared to the control group. This can be traced back either on the different sample length of the two measures or on their differences in concept and method.
As shown in Figure 2, the different sample length explains the discrepancy only partially. Still there is a remarkable difference between euro-area members and the control group for the PMR indicator.

The econometric model and its estimation
The impact of euro-area membership on the ease of doing business is estimated in a crosssection analysis. The empirical approach is basically the same as in the two previous ,2013

Econometric evidence
In this section, I present the OLS estimations of Equation (3) with the cross-sectional distance to frontier dataset. The estimates in Table 9 confirm the result of the descriptive analysis. There is no significant difference in business deregulation between euro-area countries and the control group. The coefficient of euro is insignificant in every specification. To sum up, the analysis of the DTF indicator does not confirm the hypothesis that euro-area membership induces structural reforms in business regulation. This result contrasts with the previous analysis that finds a positive impact of euro-area membership on product market deregulation. There could be several possible explanations. First, the period of observation differs between the analyses. Hence, most reforms could have been undertaken in the first years after the introduction of the euro. Unfortunately, the Doing Business indicator do not observe this important period. Indeed, as we have seen in Figure 2, the different sample length explains the discrepancy only partially. Still there is a remarkable difference between euro-area members and the control group for the PMR indicator. Therefore, some discrepancy is probably due to the differences in concept and method of the measures. I do not investigate the differences between the PMR indicator and the DTF indicator further, as this is not the focus of this paper. OLS with heteroscedasticity consistent covariance matrix estimation, i.e., robust standard errors (in parentheses), is performed. (iii) The sample covers 28 countries. harmonized unemployment rate in percent of the total labour force. (xiii) n.deval: number of devaluations during EMS from 1979 to 1993. (xiv) *, ** and *** indicate significance at the 10%, 5% and 1% level, respectively.

Conclusion
I selected three indicators of product market and business regulation: the economy-wide OECD product market regulation (PMR) indicator, the OECD indicator on regulation in network sectors (energy, transport and communications, ETCR) and the distance to frontier indicator (DTF) measuring the ease of doing business. The panel data analyses show robust evidence for a positive impact of euro-area membership on deregulation in product markets and network sectors. Euro-area countries deregulate significantly more than other OECD countries. The descriptive analysis indicates that the effect was more pronounced during the first years after the introduction of the euro.
I proceeded with a cross-sectional analysis on the effect of euro-area membership on the ease of doing business. However, I do not find any significant effects. This might be for two reasons: First, the period under observation starts several years later for the ease of doing business than for the product market regulation. As there is indication that most reforms occurred during the first years after the introduction of the euro, the distance to frontier indicator misses this important period. Second, the divergent results may be due to differences in concept and method of the measures. Indeed, the distance to frontier indicator is a much more narrow measure of the regulatory stance than the PMR indicator. Because of the weaknesses surrounding the ease of doing business analysis, I give more weight to the analyses concerning product market and network sector regulation that support a positive impact of euro-area membership.  (2014)