The recent controversies about the private conduct of Prime Minister Berlusconi did not highlight one crucial issue at stake: the economic costs of the scandals.
The Premier’s personal affairs are likely to result in considerable public outlays. First, there are the direct costs of the “entertainments”, by now public knowledge: fees for escort “services”, air-transportation costs of party guests, and so on. To these, one must add the presumed sums awarded to the health-care entrepreneurs involved in the escort-for-favours exchange in Puglia, now under police investigation.
But the reputation costs are likely to be much more important. When the private behaviour of the Premier openly contradicts publicly professed values (god, marriage, etc.), the prestige of the institutions is lessened (as the Clinton – Lewinsky affair demonstrates). At times of crisis, such disrepute may have a very negative impact on national cohesion (why should I accept sacrifices, when the Premier has a lot of fun?). Moreover, it may undermine the country standing in the international arena, thus inhibiting the capability of supporting national economic, strategic and military interests. The economic consequences of this disrepute may be substantial, but they are difficult to quantify.
There are other indirect costs of credibility, always borne by taxpayers, which are potentially very significant. Should doubts arise in financial markets about the suitability of the Prime Minister to fulfil his role, and, as a consequence, about the cohesion and stability of his government, the perceived solvency of the State may deteriorate. As a result, interest on public debt may rise. Now, since public debt has recently topped the 1,750 billion Euros, even a few basis point rise in yields would imply substantial costs for the tax payers.
But how can we assess the existence of a (possible) “Berlusconi-scandals” effect on spreads? As an academic exercise, I have tried to approximate such effects by considering the curiosity aroused by the events of the Premier, as measured by the intensity of Google searches (1) containing keywords such as Berlusconi, Noemi (the young “friend”) , Veronica Lario (the wife) , Tarantini (the entrepreneur), Topolanek (the ex Check Prime Minister, presumably appearing naked in a picture), Palazzo Grazioli, Villa Certosa, Patrizia D’Addario (the taped escort). Google Trends allows you to obtain such data on a weekly basis. Since Italian yields on 10 bonds are obviously affected by the international markets, German and US yields are also taken into account in the analysis (2).
The results of this exercise should be considered with great caution, since they may be affected by measurement errors (not all the recent searches for “Berlusconi” necessarily deal with scandals), by the limited number of observations available (29 weekly observations from 22 Dec.2008 to 23 July 2009), and by the possibility of “spurious” correlation (if, for example, students with long hair have better marks than those with short hair, one can harldy improve his vote by letting his hair grow).(3). That said, the evidence seems to suggest the existence of a positive and statistically significant relationship between the weekly return on Italian 10 years government debt (on the y-axis in the graph below) and the search intensity for “Berlusconi” (on the x axis)
Partial Correlation between 10 yield and “Berlusconi” (t-1) *
Taken literally (and admittedly, provocatively), the estimates suggest that should the rise in interest rates attributable to the Prime Minister’s scandal extend to the whole yield curve, and thus apply to the entire stock of outstanding debt, the additional burden for the budget would be of at least 593 million euro, (5) just over half of the funds allocated for the reconstruction of earthquake-stricken Abruzzi in 2009.
Even with all the precautions required, the question arises: can the country afford such a “dear” Prime Minister?
(1) Google Insights has recently been used, for example, by Hal Varian Hyunyoung and Choi, (2009), “Predicting the Present with Google Trends’,
(2) In the exercise, the Italian yield in Italy is explained by Germans and Americans 10 YR yields (Benchmark Bonds – Redemption yields, Source: DataStream), a lag, a linear trend, and the Google Trends variable lagged one period (week). I thank you Giulio Trigilia, of the University of Bologna, for the availability of data.
(4) The other keywords (Noemi, Villa Certosa, Tarantini, Topolanek,, Palazzo Grazioli, and Veronica Lario) are not significant, however. Google Trends does not have sufficient observations for Patrizia d’Addario and Villa Certosa.
(5) To evaluate the burden on public finances, I calculated the difference between the average search intensity on Berlusconi in the last 15 weeks relative to the average since the beginning of the year = [Berlusconi DELTA] = [1.698889-1.104095] and then I applied the smallest coefficient in the 95% confidence interval , emin = .0006077. Thus the estimated effect of minimum impact is = emin x [DELTA Berlusconi] x debt = .0006077 x [1.698889-1.104095]x1750 billion