Effects of fiscal policy in the Northern and Southern regions of Italy

Abstract

This paper contributes to a growing body of work within ‘fiscal policy studies,’ investigating the recent role of fiscal policy on the Italian economy. Using annual data collected on a regional basis, this study estimates and compares the (impact and cumulative) fiscal multipliers across the North and the South—the less developed area—of Italy. With recourse to a simultaneous equation model for the two macro-areas of Italy, it estimates the overall impact of the measures of budget consolidation policies during the period 2011–2013. Our analysis reveals that tax increases and, with a greater impact, spending cuts, hit the South harder compared to the North.

Keywords:

• Tax multiplier
• government spending multiplier
• fiscal policy

Subject classification codes:

• E23
• E62
• H22
• H24

Acknowledgements

We wish to thank Prof. P. Arestis and two anonymous referees for their important comments and suggestions. We also acknowledge the important help and suggestions, for the econometric work, of R. Paniccià and S. Rosignoli (IRPET; Florence).

Notes

1. Christiano et al. (2011) investigate the size of the multiplier in the zero-bound state, calibrating the ‘empirically plausible’ dynamic stochastic general equilibrium model originally proposed by Altig et al. (2011). Galí et al. (2007) estimate the responses of several macroeconomic variables to a government spending shock within a vector autoregression application (VAR).

2. The authors consider simultaneous estimation for the effects of province-specific characteristics and spending allocation; they introduce instrumental variables to test for significance after controlling for simultaneity bias. In their baseline specification, the growth of per capita value added of a province is regressed on the year-in-year change in per capita public investment on infrastructure in the same province, after controlling for the effects of the control variables (such as mafia-related murders and corruption) for each ith province.

3. The ‘marginal propensities method’ and the ‘aggregate leakages method’ are used in Faggian and Biagi (2003)’s empirical research on Italy. The marginal propensities method is based on the standard economic analysis for specifications of the multiplier. The ‘aggregate leakages method’ is based on the idea that the size of multipliers is inversely related to four possible sources of leakage (savings, imports, interregional imports and taxes) out of the local economic system.

4. It is along these lines that the US Federal Reserve still justifies its confidence in simultaneous equation model (SEM) modelling: see http://www.federalreserve.gov/econresdata/notes/feds-notes/2014/a-tool-for-macroeconomic-policy-analysis.html.

5. SVIMEZ is a think tank established in 1946 as a non-profit association. SVIMEZ specializes in the analysis and study of regional economies with a particular focus towards the regional social and economic development and Southern Italy.

6. The NUTS classification for the bi-partition of Italy, as set out by the European Commission (EUROSTAT), includes twelve Northern regions (we use the ‘Northern regions’ term to include both those properly defined as ‘Northern’ and the so-called ‘Central Italian regions,’ in the analyses which consider a tri-partition of the country) and eight Southern regions. The Northern regions are Liguria, Lombardia, Piemonte, Valle d’Aosta, Emilia-Romagna, Friuli-Venezia Giulia, Trentino-Alto Adige, Veneto, Lazio, Marche, Toscana and Umbria, while the Southern regions are Abruzzo, Basilicata, Calabria, Campania, Molise, Puglia, Sicily and Sardinia.

7. Where not otherwise indicated, all variables are expressed and estimated in log differences. Exogenous variables are evidenced in italics, and those relevant in the exercise of multiplier derivation in bold. Suffixes n, s, or ita below symbols for a variable indicate that these refer to North, South, or Italy Total, respectively. See Appendix 1 for legend for symbols and other Information.

8. More precisely, ISTAT’s published Regional Accounts cover the period 1995–2014 (Istat, various years). Data for the period 1980–1994 were reconstructed by the joint working group ISTAT-SVIMEZ, following the norms set in SEC 2010.

9. In this case, the reconstruction of the series over the period 1980–1994 was worked out by SVIMEZ staff only, upon information and suggestions by ISTAT.

10. While Regional Account data were reconstructed jointly with ISTAT staff, the bi-regional disaggregation for the main components of the public sector accounts required an elaboration effort by SVIMEZ alone, over the whole period considered; however, in this process, information made available by ISTAT for un-elaborated databases was utilized.

11. Other equations in the specification of nmods are available upon request.

12. Total industry exception construction.

13. The cost of labour is multiplied by incidence of the IRAP tax (a typically Italian indirect tax applied as a rate on the net value added of the enterprise).

14. A recent comparative study on industrial price dynamics states that ‘Apart from Italy, the evidence indicates the foreign prices are a function of domestic prices for these countries [Canada, Italy, UK, Germany, Japan, USA]. This accords with a priori expectation that large economies are not price takers (our italics)’ (Olive 2002, 74).

15. The mark-up at the aggregate economic level is defined as the following: mark-up=((ulc_bs/price_c)-1). For further comment on the theoretical frame adopted see Mott (2002).

16. More precisely, the values for the variable Lrun_c, lagged one period, are the residuals of the long-run relation: log(Ch)=α+βlog(disp_income)+γlog(net_wealth) (1) where β+γ=1.

In the derivation of the variable Lrun_C, the standard procedure was followed: (i) preliminary test ADF on each series in (1) to verify for unit root; (ii) the same test applied for the same variables expressed in first differences; (iii) final test for the existence of a cointegrating relationship among the series (Johansen’s cointegration test). Such tests for validation of the procedure are reported in tables, 2B and 2C in Appendix 2, respectively.

17. This variable should take into account an opportunity cost in the consumption/saving choice. see Pozzi (2015).

18. The presence of foreign tourists in the South accounts for only 13% of the total value of the country.

19. On average, the public component accounted for 23% of total consumption in the North and 31% in the South.

20. By comparison internationally, Italy shows average dimensions of firms among the smallest. In the branches of ‘traditional’ manufacturing (in the taxonomy à la Pavitt), which covers the majority of the Southern firms, the characteristic dimension of the firm (entropic average) is about 60% of the value of the North.

21. We believe that this ratio is more suitable, with respect to absolute volumes, in detecting the phenomenon of a ‘credit crunch’ as significant after the recent financial distress.

22. In the North, construction accounts for less than half of total investments; in the South, the ratio rises to about 60%.

23. The demand-side block is closed by the equations for regional exports and imports (national), not reported here.

24. For full discussion of the aggregate supply function within the NMODS frame, see Paniccià et al. (2011) and (2013). This has a macroeconomic foundaton in Kaldor (1957).

25. The meaning of this inclusion, we believe, is then different with respect to the considerations earlier made for the case of machinery investment. In this case we are not able to separate, within the whole values of loans _ind , a component destined to increase investment capacity from that financing cash flow for the current operation; a single variable was thus used as a proxy for both.

26. The share of industrial value added on regional GDP is only 8% in the South, compared to 18% in the North. The average dimension of foreign-participated firms (about 130 employees per plant) in the South is almost 25-fold higher than that of the average local production unit.

27. The extent to which firms set the price is as follows: P_i = C_i + αC_i where P_i is the price of the firm i, C_i the full cost, and α is the margin over full cost, and if α depends upon the firm’s higher technological capability, firms located in the North may target a higher price compared to firms located in the South. This means then that an increase in cost due to the taxation might result in a price increase that is larger in the North than in the South over the short term. Other research within SVIMEZ reveals that the South accounts for about 30% of all typologies of work that might be defined as ‘atypical’ or ‘informal’ (occasional dependent and non-dependent work, etc.); this implies that the influence of traditional factors behind price fluctuations in the more developed economies, such as unit cost of labour calculated from official indexes of labour remuneration and productivity, might have, in the immediate term, a smaller incidence in determination of the fluctuations of consumer prices.

28. Non-linearity of the model is derived mainly from the fact that structural relationships are estimated in logarithmic differences; this implies that the reduced form, from which the multipliers are calculated, will have coefficients that are variable among reference periods.

29. This can be evidenced in more detail by the analysis of the input–output matrices for the North and the South: the latter is, in fact, more ‘empty’ for many intersectoral transactions (Cherubini et al. 2011).

30. We offer as a reminder that clauses for increases in VAT, or other indirect taxes, are often imposed as a ‘default clause’ in the consolidation plans created in agreement of supranational bodies (EEC Commission, IMF), when ‘spending review’ or other targeted policies fail to meet their quantitative targets.