The Eurozone Recession: Are There Alternatives?
Product Markets: The Evidence
The Data on Regulation
We use yearly data on twenty-one OECD countries (Australia, Austria, Belgium, Canada, Switzerland, Germany, Denmark, Spain, Finland, France, the United Kingdom, Greece, Ireland, Italy, Japan, the Netherlands, New Zealand, Norway, Portugal, Sweden, and the United States), covering a maximum time span from 1975 to 2003. The data come from a variety of different sources.
We use time-varying measures of regulation for seven nonmanufacturing industries in twenty-one OECD countries for the period from 1975 to 2003. The data have been collected by Conway and Nicoletti (2007) from both national sources (by means of specific surveys) and published sources and are described in detail by Nicoletti and Scarpetta (2003). The regulatory indicators measure, on a scale from 0 to 6 (from least to most restrictive), restrictions on competition and private governance in the following industries: electricity and gas supply, road freight, air passenger transport, rail transport, post, and telecommunications (fixed and mobile).
The summary index of regulation includes information on entry barriers, public ownership, the market share of the dominant players (in the telephone, gas, and railroad sectors), and price controls (in the road freight industry). Entry barriers cover legal limitations on the number of companies in potentially competitive markets and rules on vertical integration of network industries. The barriers to entry indicator takes a value of 0 when entry is free (i.e., a situation with three or more competitors and with complete ownership separation of natural monopoly and competitive segments of the industry) and a value of 6 when entry is severely restricted (i.e., situations with legal monopoly and full vertical integration in network industries or restrictive licensing in other industries). Intermediate values represent partial liberalization of entry (e.g., legal duopoly, mere accounting separation of natural monopoly and competitive segments). Public ownership measures the share of equity owned by central or municipal governments in firms of a given sector. The two polar cases are no public ownership (a value of 0 for the indicator) and full public ownership (a value of 6 for the indicator). Whenever data are available (i.e., telecoms, air transport), intermediate values of the public ownership indicator are calculated as an increasing function of the actual share of equity held by the government in the dominant firm. In some cases (e.g., the energy industries), a simpler scale is used, pointing to full or majority control by the government (a value of 6), various degrees of mixed public/ private ownership (intermediate values), and marginal public share or full private ownership (a value of 0).
The construction of the indicators by the OECD involved the following steps. First, they separated indicators for barriers to entry, public ownership, and market share of new entrants, and price controls were created at the finest available level of industry disaggregation (e.g., mobile and fixed telephony). Second, they aggregated indicators at the industry level, taking simple averages or revenue- weighted averages (when aggregating horizontal segments of industries, such as mobile and fixed telephony). Third, they computed the index of overall regulation by averaging, in each of the seven industries, the indicators of barriers to entry, public ownership, market share of new entrants, and price controls.
Here, we used simple averaging of the indices to reach the level of industry aggregation for which macroeconomic data (value added, labor costs, and employment) are available. More specifically, we have aggregated the regulation indices for the seven sectors in three broader sectors: energy (electricity and gas), communication (telecommunications and post), and transportation (airlines, road freight, and railways).
In our benchmark regressions, we use the regulatory indicator REG, which includes all dimensions except public ownership. In the sensitivity analysis, we also consider three other indicators of regulation: the overall indicator, including all the regulation dimensions; one indicator that summarizes barriers to entry (comprising legal restrictions and vertical integration); and one indicator that includes only public ownership information.
In the augmented regressions, we introduced two additional sectors: retail and professionals. Data on regulation in these two sectors in twenty- one OECD countries are available only for two years: 1996 (for professionals) or 1998 (for retail) and 2003. These regulatory indicators range from 0 to 6 (from least to most restrictive). In the retail sector, they capture three components: barrier to entry, operational restrictions, and price control. For the professionals, indicators measure entry regulations and conduct regulations in four sectors: accounting, architecture, engineering, and legal services. For a detailed description, see Conway and Nicoletti (2007).
The Macroeconomic and Political Data
The economic data on value added, labor costs, and total employment at the country- sector- year level for the period from 1975 to 2003 come from the OECD STructural ANalysis (STAN) database for industrial analysis, revision 3 (ISIC rev. 3). This database covers both services and manufacturing sectors for the OECD countries. The macroeconomic data for the nonmanufacturing sectors for which we have indices of regulation are available at the following level of industry aggregation: (a) electricity, gas, and water; (b) communications and posts; and (c) transport and storage. From now on, we will name the sectors defined in (a), (b), and (c) as energy, communications, and transport, respectively. We merge the data from the STAN data set with the database containing the regulation indices. As mentioned previously, because data on value added, labor costs, and total employment are not available for each single industry for which regulation indices exist, we mapped the industry- level regulatory indicators into the nonmanufacturing aggregates covered by the STAN database.
Macroeconomic data at the country- year level are from the OECD Economic Outlook number 80 database. Finally, the Database of Political Institutions (DPI) of the World Bank, compiled by Beck et al. (2001) and updated in 2004, contains all the political variables employed in the analysis.
Patterns of Product Market Deregulation
Beginning in the late 1970s, OECD countries have initiated a broad- based process of deregulation. They were not all starting from the same initial position, however. Generally speaking, Anglo- Saxon countries (the United States, in particular) were less regulated than continental European countries, and they started to deregulate early: the United States and the United Kingdom in the early 1980s, New Zealand in the late 1970s, and Ireland in the late 1980s. In the last two decades, there has been convergence: the difference in the degree of regulation of product markets (at least for the sector for which we have data) is lower now than it was in the early 1980s. The laggards are catching on.
In what follows, we divide the countries into three groups: (a) those that adopted the euro (the EMU group); these countries are Austria, Belgium, Finland, France, Germany, Ireland, Italy, the Netherlands, Portugal, and Spain; (b) those that are part of the European Union but did not adopt the euro (the European single market group, or ESM); these countries are Denmark, Sweden, and the United Kingdom; and (c) those that are not in the European Union and obviously do not have the euro; these countries are Australia, Canada, Japan, New Zealand, Norway, Switzerland, and the United States.
All sectors have deregulated-communications more than any other and energy less than any other. Non- EU countries have deregulated less, but as we said before, they were starting from a much lower average level of regulation. The single market group has deregulated most, but in the period from 1999 to 2003, the EU countries have picked up momentum, having done very little until then, especially given their high initial level of regulation. With the exception of Ireland, very few EU countries did much in terms of deregulation in the 1980s, so leaving Ireland out, the pattern for the EU countries would be even more skewed toward the recent period. The ESM group includes the United Kingdom, which started deregulation early, as did other English-speaking countries, and also includes Nordic countries, which have deregulated quite a lot and performed some pattern of convergence in the deregulation process: since 1999, the countries that deregulated more were clearly those that had higher degrees of regulation until the mid- 1990s.
All our regressions are estimated with generalized least squares, allowing for heteroschedasticity of the error term; they include the lagged value of the left- hand side variable, as well as country, sector, and time dummies. Sensitivity analysis confirms that all the results are robust to controlling for country- sector specific dummies, time trends, and country- specific time trends.
We estimate also our basic specification of the level of regulation (measured by the indicator variable REG). We included data on the three sectors of transportation, energy, and communications; also included the two additional sectors: retail and professionals. We measure the impact of the single market program and of the euro on regulation with the dummy variables ESM and EMU. Specifically, ESM is an indicator variable equal to 1 from 1993 onward for all countries that belong to the European Union (i.e., Austria, Belgium, Denmark, Germany, Finland, France, Greece, Ireland, Italy, the Netherlands, Portugal, Spain, Sweden, and the United Kingdom) and equal to 0 otherwise.
The indicator variable EMU is equal to 1 from 1999 onward only for those countries of the European Union that have adopted the euro (i.e., Austria, Belgium, Germany, Finland, France, Greece, Ireland, Italy, the Netherlands, Portugal, and Spain) and equal to 0 otherwise. Both the single market and the euro have accelerated deregulation: the coefficients of ESM and EMU are negative (equal to – 0.064 and – 0.18, respectively) and statistically significant at the 5 percent level or better. Interestingly, the adoption of the euro has had a larger (about three times as large) impact on regulation than that of the single market program, and for a country that participated in the single market and adopted the euro, our estimates imply that the level of regulation decreased by about – 0.25 points. We check whether these results hold for each sector. The adoption of the euro was especially important for energy and communications, while the single market was key for transportation and had no statistically significant effect in the energy and communications sectors.
Finally, we investigate whether the effect of the single market program and the adoption of the euro depend on the initial level of regulation by adding the variables ESM. The effect of the single market is independent of the level of regulation: the coefficient of the interaction term between the single market dummy and the level of regulation lagged one is not statistically significant, both in a specification in which we exclude the
variable EMU∗REG (–1) and in one in which we include it. (Results are not shown but are available upon request.)
On the contrary, shows that the effect of the euro was larger when the initial level of regulation was larger, reemphasizing the process of convergence mentioned previously. The coefficients of the dummy variable EMU in the energy and communication sectors become positive but insignificant. However, the magnitude of the coefficients of the variables EMU∗ENERGY and EMU∗COMMUNICATION and of EMU∗REG (–1) imply that for each value of REG (– 1) observed in the energy and communications sectors, adopting the euro is always associated with deregulation.
(1) In which the two additional sectors, retail and professionals, are also included. The estimates show that the single market, not the euro, was important for the retail sector and that the professionals sector has not been deregulated at all.
Finally, the regulatory variable that we are using (REG) looks at all aspects of regulation, except the one of public ownership. Results hold when we use the indicator of regulation that only measures barriers to entry and vertical integration and the more general indicator that also looks at public ownership.
Summarizing, the introduction of the euro has contributed to structural reforms in the product markets. This effect is above and beyond the effect of membership in the European Union from 1993 onward. Moreover, deregulation was stronger in EMU country- sectors with higher initial levels of regulation. This may give some prima facie and indirect support to the idea that deregulation was most needed once countries could not rely on exchange rate devaluations to boost competitiveness. In fact, the more heavily regulated (and less productive and competitive) country- sectors may have been those suffering the most from the loss of competitive devaluations and hence the ones that were forced to liberalize the most.
.Why Should the Euro Matter? Empirical Evidence
One of the reasons why a country joining the EMU may want to adopt structural reforms is that the competitive devaluation channel is not available anymore as a tool (or a palliative) to regain competitiveness.15 Lacking competitiveness indicators at the country- sector year level for the period from 1975 to 2003 for the energy, communications, and transport sectors, we measure competitiveness with variables varying only along the country- year dimension. We use two different indicators: the growth rate of the Consumer Price Index (CPI) relative to competitors at t – 1-COMPET1 (– 1)—and the growth rate of the export goods deflators relative to competitors at t – 1-COMPET2(– 1). We include the linear and quadratic terms to capture for possible nonlinearities; we add the interaction term of the competitiveness indicators and the EMU dummy variable to investigate whether the loss of exchange rate devaluation as a policy instrument to boost competitiveness leads to structural reforms. The coefficients of the variables COMPET1 (– 1) and COMPET2 (– 1) and their squares are not statistically significant at conventional critical values, suggesting that deregulation reforms do not generally occur in countries that are losing competitiveness. However, this is not true for countries that adopted the euro. In fact, the interaction terms of the competitiveness indicators and the EMU dummy variable are negative and statistically significant at the 5 percent level, suggesting that for EMU countries, the higher the growth rate of CPI and export goods deflators relative to competitors at t – 1, the larger the decrease of the regulatory index. Finally, we control for the number of devaluations those countries that adopted the euro experienced in the period from 1979 to 1993. Our idea is that only countries that de facto used the exchange rate as a tool to regain competitiveness should suffer from its loss and liberalize markets. The variable N. OF DEVALUATIONS FROM 1979– 1993 is equal to 5 for France, 1 for Belgium, 7 for Italy, and 3 for Ireland. It is equal to 0 otherwise. For the EMU countries, the more devaluations a country did from 1979 to 1993, the larger the decrease of the regulatory index (but the coefficient is statistically significant only at the 10 percent level).
Two caveats are worth mentioning. First, we are treating our competitiveness indicators as exogenous. While this clearly may not be the case, note that here, we are not really interested in the effect of competitiveness on regulation but instead on its differential effect among EMU and other countries. Hence, even if the competitiveness indicators were not exogenous, it is not clear why the bias in our estimates should differ among EMU and other countries. Second, the coefficient of the variable EMU∗REG(– 1) remains negative and statistically significant, as in table 2.1, suggesting that: (a) our competitiveness indicators are not capturing the loss of competitiveness, and hence the need of reforms, very well when the exchange rate instrument cannot be used anymore; (b) the euro is important for structural reforms in product markets for other reasons beyond the fact that the competitive devaluation channel is not available anymore; (c) what we are identifying as a euro effect is just picking up the impact of some omitted variable; and (d) any combinations of (a), (b), and/ or (c).
Other Determinants of Product Market Reforms
We also check that accounting for other critical elements that drive reforms does not alter the results we discussed so far on the effect of the euro on the deregulation of product markets. We begin by testing whether various variables that measure the macroeconomic conditions of each sector matter. We include the sectors’ value added, labor expenses, and total employment at time t – 1, measured as a share of country’s total value added, labor expenses, and total employment at time t – 1. Blanchard and Giavazzi (2003) suggest that in the short run, product markets’ deregulation reforms generate costs for both incumbent firms and their workers. Hence, incumbents tend to oppose such reforms. When rents are lower, however, resistance to deregulation falls, as the incumbents’ short- term losses can be easier outweighed by the future benefits of deregulation. Results support this argument.
In fact, we find that regulation decreases when value added and labor costs of the sector fall-that is, when the sector’s rents decrease. We also find that product markets are deregulated in country- sectors- years with lower employment. Hence, in less labor- intensive sectors, governments can meet less resistance and can more easily implement deregulation measures. We also investigate whether there are differential effects between EMU and non- EMU countries relative to the effects of value added, labor costs, and employment on regulation, but on this score, we found no differences between EMU and non- EMU countries. We augment the specifications with several macroeconomic and political controls. We investigate the crisis hypothesis, the role of the countries’ fiscal conditions, the timing of reforms in relation to the electoral cycle, the interaction between reforms in the product and labor markets, and the effect of reforms occurring in trading partners’ countries. All variables are measured at time t – 1, both to allow for the fact that it may take some time until governments react to macroeconomic events and to reduce the possibility of reverse causality in our estimates. Several results are worth noting. First, the results on EMU shown thus far are robust to the inclusion of the additional control variables. Second, we find evidence that deregulation reforms occur in country- years in which the output gap (defined as the difference of actual output to potential) is below the ninetieth percentile of the output gap empirical density (equal to – 3.4 percent). This gives some support to the crisis hypothesis-namely, that reforms are more likely to occur in bad times. Third, the higher the primary deficit as a share of gross domestic product (GDP), the lower the level of regulation, indicating that reforms’ blockers may be less powerful when they feel that public finances are also in trouble and that liberalizing the economy can help both in boosting growth and possibly in reducing the likelihood of further increases in taxes or cutting in spending. Fourth, we find some evidence that product market reforms happen at the beginning of the political term (right after an election), but this result is not particularly robust to specification changes. Fifth, deregulation in trading partners fosters deregulation at home.
Finally, we looked into the interaction between labor market reforms and product market reforms. Specifically, our estimates show that an increase in unemployment benefits leads to lower regulation in product markets, while a decrease in the employment protection index is associated with less regulation of product markets (but the coefficient is significant at the 10 percent level only. Product market liberalization reforms seem easier to implement if workers receive some kind of protection in the form of social insurance. As mentioned earlier, workers of the incumbent firms are more likely to become unemployed and lose in the short run from deregulation.
Hence, they can be more willing to bear the short- run costs once the generosity of unemployment benefits increases than they otherwise would. Fiori et al. (2007) find that labor market reforms do not Granger- cause product market reforms. However, their labor market indicator is the principal component of unemployment benefits and employment protection. Results show that the two variables have opposite effects on regulation in product markets. Hence, considering a combination of the two variables may prevent one from detecting any effect of labor market regulation on product market regulation.
Endogeneity of Euro Membership
The decision to join the EMS and especially to adopt the euro is of course not an exogenous variable. In order to investigate this issue, we have estimated using an instrumental variable procedure. First, we have estimated with a probit model the probability that a certain country adopts the euro. The choice of the right- hand side variable is based upon the gravity literature on trade and the literature on currency unions.16 The specification, described in detail in Alesina, Ardagna, and Galasso (2008), is meant to capture that: (a) countries that trade more with each other should be more likely to choose to be part of the same common currency area; (b) the higher the correlation of the business cycle frequency (output and prices), the more likely it is that two countries will choose to join the union; and (c) the higher past inflation, the more likely it is that a country will join the union. In fact, the more two countries trade with each other, the more they benefit from a common currency. The more correlated are their business cycles, the lower the costs of a simple monetary policy. Finally, a history of high inflation makes a monetary anchor especially effective. We find support with regard to EMU for the fi rst two effects but not for the third.17 This is not surprising, as the monetary anchor argument certainly did not apply to low- inflation members (e.g., Germany and France).
We then use the estimated probability of joining the union as an instrumental variable. The results, shown in Alesina, Ardagna, and Galasso (2008), indicate that the coefficients of interests on EMU are generally robust to this procedure. We have investigated all the specifications with various degrees of success. In some cases, the results remain significant, while in some cases, the standard errors are too big for statistical significance. We are not convinced that the decision of whether to enter the euro area was exogenous only (or mainly) to economic variables. Political consideration seemed crucial, and therefore it is hard to measure with an instrument the decision of whether to join.
14. We also checked whether the countries that deregulated after the adoption of the euro in the years following 1999 had experienced a delay in deregulation because they were too busy achieving the target criteria to join the monetary union. More specifically, we tested what happened to EU countries in the run- up to the euro during the period from 1993 to 1999. We did not find any evidence of an effect of postponement.
15. Ssome microeconomic evidence suggesting that sectors that have gone through deeper transformations and that enjoyed more productivity gains are exactly those that benefited more from pre- 1999 devaluation.
16. See Alesina, Barro, and Tenreyro (2002), in particular.
17. Also, Rose (2000) fi nds a signifi cant and negative impact of the inflation rate on the probability of joining a currency union.
