![Sample our Economics, Finance,Business & Industry journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5931/TOC-Subject-Sample-2021-Economics-Finance-Business-Industry.jpg"})
Abstract
According to the literature, the decline experienced by the Italian economy in the last two decades depends on a slowdown of its labour productivity, starting in the 1990s. The supply-side explanations of this slowdown are inconsistent with the major stylised facts. In this paper, we verify whether a better explanation is provided by the effect of a negative demand shock, through Italy’s external constraints, in the framework of Kaldor-Dixon-Thirlwall’s cumulative growth model. To this end, we use a multi-country generalisation of Thirlwall’s balance-of-payments-constrained growth model, which allows us to investigate the contribution of Italy’s main trade partners to Italy’s long-run growth from 1970 to 2010. The trade partners are disaggregated into seven groups: Eurozone core, Eurozone periphery, United States, other European countries, OPEC countries, BRIC, and the rest of the world. The results show that Italy’s long-run growth has been consistent with the Bop-constraint, that its decline can be explained by a progressive tightening of this constraint, that the sudden slowdown of labour productivity in the 1990s corresponds to a major shock on Italy’s external constraint, and that the major contributions to this shock came, through different channels of transmission, from the core Eurozone countries and from OPEC countries.
Acknowledgements
The author thanks Elias Soukiazis and two anonymous referees for their helpful comments, as well as Christian Alexander Mongeau Ospina for his research assistance, and Patrick Lynch for revising my English. All remaining errors are my own. Financial support from the Nando Peretti Foundation (Grant 2014-07) is gratefully acknowledged.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes
$ Paper presented at the 15th INFER Annual Conference (University of Orléans, 29 May – 1 June 2013).
1. On 1 January 1999, the euro was adopted as an official currency by the 11 countries that had met the so-called Maastricht criteria: Austria, Belgium, Finland, France, Germany, Ireland, Italy, Luxembourg, the Netherlands, Portugal, Spain. In 2000, Greece qualified and was admitted to join the Eurozone on 1 January 2001 when the euro replaced the national currencies in circulation.
2. Following Busetti et al. (Citation2006), we identify as the EZ ‘core’ the countries that belong to the same ‘inflation convergence club’ as Germany (i.e., Austria, Belgium, Finland, France, Luxembourg, and the Netherlands), and as ‘periphery’ the high inflation countries (Greece, Ireland, Portugal and Spain), whereas Italy constitutes a ‘medium inflation club’ in its own.
3. Since the BoP constraint is derived under the hypothesis of current account equilibrium, a real exchange rate depreciation relieves the constraint to the extent that the Marshall-Lerner condition is met. It is worth noting, however, that whenever the current account starts from a deficit position, more restrictive conditions are needed. Gandolfo (Citation2002, par. 7.2) provides a full discussion.
4. I thank an anonymous referee for this remark. While limiting its usefulness for the purposes of economic policy analysis, this shortcoming leaves unaffected the capability of the model to explain the historical developments of the BoP-constrained growth rate, conditional on an historically given path of nominal wages. The purpose of the original paper by Dixon and Thirlwall (Citation1975) was to explain regional growth differentials. This justified the assumption of nominal wage exogeneity, since it can be argued that collective agreements at the national level make the evolution of nominal wages largely exogenous to region-specific features.
5. More precisely, and
, where Xj are exports to partner country j, Ej is the bilateral nominal exchange rate (units of reporting country’s currency for one unit of partner’s j currency), Pj are country j export prices, and Mj are imports from partner j.
6. In the limit case in which n=1, the import and export trade shares are equal to 1 and equation (3) reduces to the aggregate formulation of Thirlwall’s law (the first equation of model (1) above).
7. Consider that in the import functions, income elasticities are usually larger than price elasticities.
8. By summing-up these conditions over the n partners we get the aggregate Marshall-Lerner condition.
9. The split of the EZ into three ‘clubs’ has been discussed in Section 2 above.
10. Since the ITCS database starts in 1988, its series were retropolated back to 1970 using the rates of change of the corresponding time series provided by the 2008 version of the CHELEM database (CEPII Citation2008), that reports data from 1967 to 2007. The homogeneity between the two datasets was tested by calculating the correlation coefficient between the rates of change of each couple of corresponding series in the overlapping sample (1988–2007). We considered the rates of change to avoid spurious correlation issues. The estimated correlations range from a minimum of 0.772 (exports to the BRIC) to a maximum of 0.995 (imports from OPEC), with an average of 0.92. The data provided by the two databases are therefore substantially homogeneous, which is to be expected since they are constructed starting from the same national sources.
11. When repeated with differenced variables, the tests consistently rejected the unit root null, which implies that non-rejection for the series in level means presence of a single unit root. Detailed results are available upon request.
12. Gregory and Hansen’s procedure allows also for the presence of a shift in the intercept only (from α0 to α0+α1). Whenever the non-cointegration null was rejected against this ‘intercept shift’ alternative, we preferred to adopt it, as it resulted in a more parsimonious parameterisation of the model.
13. Consider, for instance, the income elasticity of exports to partner 4, namely ε4. This parameter changes after 1985, going from 3.13 to 1.70 (Table ). As a consequence, the corresponding ‘partner growth’ effect in the numerator of equation (3), ν4ε4, was evaluated as 0.120×3.13×0.031=0.012 up to 1985 (where 0.120 is the average exports market share of ‘other European countries’ and 0.031 is the average real growth rate from 1970 to 1985 of these countries) and as 0.124×1.70×0.021=0.004 from 1986 on (where 0.124 and 0.021 are the average exports market share and real growth rate from 1986 to 2010).
14. Following Ravn and Uhlig’s (Citation2002) suggestion, the degree of smoothness λ of the Hodrick-Prescott filter was set equal to 6.25. As for the Christiano-Fitzgerald filter, its asymmetric version was used to remove those cyclical components that have a period of oscillation between 2 and 8 years (following the standard definition of business cycle that goes back to Burns and Mitchell Citation1946).
15. The respect of the Bop constraint does not imply, per se, that a country’s growth equals that of its productive potential. In fact, actual growth may respect the constraint, while being below the growth of productive potential (Thirlwall Citation2001). I thank an anonymous referee for this observation.
16. Since the income elasticity of exports towards EZ peripheral countries underwent a change in 1986 (Table ), its average increased by 0.14 between the first and the second subsample.
17. It is worth remembering that under the EMS exchange rate mechanisms, the realignments preserved some degree of symmetry. Indeed, during the 1980s the Deutsche Mark was realigned upward six times, and the Italian lira was realigned downward only four times (also owing to the fact that it had negotiated a wider oscillation band around the declared central parity with the ECU).