Birds of a feather flock together: trust in government, political selection and electoral punishment

According to Uslaner (2008, p. 290), “generalized trust represents a sense of social solidarity, a belief that other people, especially people unlike yourself, are part of your moral community”.

As explained by Golden and Chang (2001), Italy represents an extreme case of political corruption in an established democratic setting. The evidence that wealth does not account for perceived corruption should be considered as an exception to be included in any theory of political corruption. Italy offers a unique source of data on political corruption in terms of RAPs. Owing to the “Clean Hands” investigations since 1992, Italy offers one of the most important single sources of information about the nature, extent and functioning of general political corruption in a modern democratic system.

For a detailed review of the results on empirical indicators of social capital, see Grootaert and van Bastelaer (2001).

Bjørnskov and Sønderskov (2013) revived the relevance of social capital, confirming that it is comprised of multiple dimensions.

We also test with the single measure of cheating or standardize by its standard deviation. The empirical findings are similar to those reported here.

The measure of constraints on the executive was coded for different 40-year windows around the years 1600, 1700, 1750, 1800 and 1850; it takes values from 1 (unconstrained authority) to 7 (maximum accountability and constraints).

Unlike Chang et al. (2010), we do not consider the XI (1992) and the XII legislatures (1994) since the political scandal of “Clean Hands”, that took place in those two legislatures, would impart bias to our estimates.

Furthermore, Chang et al. (2010) report that, differently from the XI (1992) and the XII legislatures (1994), from 1948 to 1987 the average reelection rate of politicians already involved in corruption scandals was closer to the reelection rate of other politicians. That evidence minimizes the potential risk of politicians’ selection bias.

In fact, the totality of political parties and a large fraction of the politicians active in the 1948–1987 period no longer are present in the 2007–2009 political scenario.

In the "Appendix", Table A.1 supplies details on variable definitions and data sources, whereas Table A.2 reports descriptive statistics.

The electoral rule for seats in the Italian House of Representatives between 1948 and 1987 was a pure proportional system. Thus, deciding to normalize the RAP and S_RAP variables for inhabitants seems to us the correct choice.

In the observed time span, we have 32 electoral districts.

As far as we know, in Italy, blood donations are done on a voluntary basis and do not entitle blood donors to receive any amount of money.

In their "Appendix" the authors check for alternative measures of social capital based on principal-component analysis.

For convenience, in the following estimates we report only the variables of interest. The full estimates are available from the authors upon request.

The variable was coded for different 40-year windows around the years 1600, 1700, 1750, 1800 and 1850; it takes values from 1 (unconstrained authority) to 7 (maximum accountability and constraints). For more details, see Tabellini (2010).

To provide a comparison with the results obtained by Nannicini et al. (2013), we also report in "Appendix A" the estimates using blood donations as a measure of social capital. Specifically, in each column of Tables A.5 and A.6, we estimate a baseline model analogous to the baseline estimates in column 1 (using blood donations) of Table 4 and in column 4 (using the cheating index) of Table 5. The only difference is the exclusion of dummy variables for macro geographical areas (i.e., North-West, North-East, Center, South, Islands), owing to the large correlations of those geographical areas with our instruments.

Incidentally, average per capita blood donations as measure of social capital seems to suffer endogeneity in explaining politicians’ misbehavior.

We also test potential endogeneity using as instruments the Institutional Quality Index (Nifo and Vecchione 2014) and the European Quality of Government Index (Charron et al. 2014), with results analogous to those reported here. The results are available from the authors upon request.

For convenience, in the following estimates we report only the variables of interest. The full estimates are available from the authors upon request.

As a further robustness check and to offer an additional comparison between the two measures, in "Appendix A" we also report the estimates for two subsamples based on the median of blood donations and trust in government, respectively (Nannicini et al. 2013). Operationally, we split the sample by pooling the observations lying above or below the median values of blood donations and trust in government. Specifically, in both Tables A.7 and A.8 columns 1 to 4 represent the estimates of the split samples based on the median of blood donations in the electoral districts. Similarly, columns 5 to 8 report the estimates of the split samples based on the median of our measure of trust in government at district level. Tables A.7 and A.8 show that the incidences of RAP or S_RAP cause electoral punishment of politicians involved in misbehavior in the electoral districts exhibiting high levels of trust in government. Incidentally, the magnitude of the coefficients on electoral punishment seem slightly smaller than in the estimates based on blood donations.