Discussion of the paper by G. Albareto et al., "The real and financial wealth of Italian households by region"

The paper by Albareto et al. puts forward new estimates of the real and financial wealth of Italian households, broken down by region, in the years 1998-2005. It forms part of a thirty-yearlong project by the Bank of Italy to estimate these aggregates. More recently, Bank researchers have paid special attention to the geographical distribution of wealth. The work of Albareto et al. offers two major improvements over the preceding literature. First, the real and financial aggregates include components formerly excluded. Second, the estimates use a new series of house prices, correcting earlier overestimation. The authors revise the value of the real component of wealth significantly downward, most sharply for the Centre and the North of Italy, somewhat less so for the South for the three parts of the country. It is worth underscoring that this adjustment occurs notwithstanding the inclusion of the new components mentioned. The reduction in the value of households’ real wealth at current prices depends on the use of a more accurate house price index than that used in Cannari, D’Alessio and Paiella (2006). I shall discuss some of the paper’s most interesting conclusions. The new estimates show the substantial regional disparities in net per capita real and financial wealth. The authors effect a comparison, for 2004, with the estimates of Cannari, D’Alessio and Paiella (2006), which they take as benchmark. Setting the estimate for the North equal to 100, Cannari et al. (2006) get values of 89 for the Centre and 44 for the South. For the same wealth components, Albareto et al. get values for Centre and South of 87 and 50 respectively of that for the North. That is, they diminish the weight of the Centre (from 89 to 87) and increase that of the South (from 44 to 50). The new estimates, moreover, produce a different ranking of the regions according to per capita real wealth. The paper confirms the findings of Cannari et al. (2006) on the sharp geographical disparities within Italy, but with some differences that should be emphasized and fully brought out. For one thing, the disparity between North and South is attenuated. Second, the regional ranking is changed. This is an intriguing point, bringing to mind a question that I should like to put to the authors. The comparison with the data in Cannari et al. (2006) is only for 2004 and thus indicates just the “level” of the discrepancy in that year. It might be interesting to extend the comparison to the entire period that the two works have in common, i.e. 1998-2004, so as to get an idea of the trend in the discrepancy in the values of the real and financial components of wealth over time. In fact, in terms of economic analysis, the process of asset revaluation has significant impact on spending and saving decisions, and the dimensions of the process are significant during this period. 2. The geographical distribution of GDP as criterion for liquidity assignment I was particularly struck by the magnitude of one aspect of the geographical distribution of which my perception had been what you might call “qualitative”. This is the North-South gap in the weight of real as against financial assets, risky as against safe financial assets, and liquidity (cash and postal deposits in particular). There is a substantial literature on this point explaining the importance of real as against financial assets in the South, and among the latter the greater weight of safe and specifically liquid assets. The determinants generally mentioned include the South’s lower level of total wealth, greater perceived uncertainty about future incomes, low levels of social capital, and so on (Guiso, Jappelli, Sapienza and Zingales, which the paper cites). In addition to the social, demographic and economic characteristics of the population, other relevant factors involve the supply of banking services (fewer branches and e-money terminals) which are said to weigh on households’ financial choices in the South. The possible causes of the differing composition of real and financial wealth in the South by comparison with the Centre and North are numerous. Inquiry in this field therefore demands the use of indicators targeted at the territorial distribution of financial items that are far from the Bank of Italy’s supervisory tools and from prudential reporting requirements: this applies in particular to holdings of banknotes. The procedure adopted by Albareto and his co-authors to disaggregate nationwide holdings of banknotes on a regional basis consists simply in following the distribution of GDP. This criterion, in fact, is standard in the Bank’s analyses, such as the works of Magnani (1997) and Berrettoni et al. (1999). As I see it, using GDP as the gauge of the distribution of cash in circulation could well produce an overestimate of the figure for the Centre and North and an underestimate for the South, although the paper’s reference to households as consumers and as producers is an argument in favour of that approach. In my view, the bottlenecks in the supply of financial services – indicated by the importance of postal deposits in the South (the geographical attribution of these assets being reliable) – together with the economic, social and demographic factors mentioned above, counsels special care in specifying the factors relevant to the geographical assignment of liquidity (Arciero et al, 2006, p. 312). I wonder whether we might not do well to consider, in addition to GDP, such things as the share of the illegal and underground components of total output, the diffusion of ATM and POS terminals, and the number of bank and post office branches. I am well aware that in recent years the item “banknotes” does not weigh very heavily in total financial wealth at national level. At the end of 2005 they represented about 2 per cent of total financial assets in Italy, but this rises to 2.7 per cent in the South and Islands. Given the relatively low level of this item nationally, a different geographical distribution of holdings of banknotes based on alternative criteria is not likely to make a significant contribution, unless there is very marked polarization (which cannot be ruled out in advance). 3. A matter of editing, and of substance It is important for applied economic researchers to be apprised immediately, in a publication, of the definitions used for macroeconomic variables and the way in which they are constructed statistically. Obviously there are excellent reasons for placing this information in compact form in a statistical or methodological appendix, but this common practice may lead less attentive readers, or those less directly involved in that aspect of the study, into error. Further, there is no denying that the statistical data made available always reflect a mix of the theory that demarcates their content and the operative choices that determine their practical definition in producing statistical estimates. The point here is that Albareto and his colleagues should shift their account of some of the criteria used in determining the geographical distribution of real and financial aggregates from the appendix to the main body of the text, at least in broad outline. 4. Three brief suggestions In closing let me express my great appreciation for this essay, which will be the basis for interesting further developments in various branches of economic analysis. This 5 appreciation extends to the other researchers here at the Bank who have made and are still making contributions in this sphere. To sum up, I think the authors of this work could improve it in three ways: 1. by extending the comparison of their results with those of earlier works (in particular Cannari, D’Alessio and Paiella, 2006) to all the years for which this is possible; 2. by adopting other criteria in addition to GDP to determine the distribution of banknotes in circulation, an item that is of modest size nationally but substantial in some parts of the country; 3. by including a discussion, possibly in the main body of the text, of the most interesting implications of their criteria for the geographical distribution of wealth, which is now found in the methodological appendix.