What Predicts Corruption?

Abstract

Using rich micro-data from Brazil, we show that multiple machine learning models display high levels of performance in predicting municipality-level corruption in public spending. Measures of private sector activity, financial development, and human capital are the strongest predictors of corruption, while public sector and political features play a secondary role. Our findings have implications for the design and cost-effectiveness of various anti-corruption policies.

Corruption and complexity: a scientific framework for the analysis of corruption networks

“…The possibility of establishing models that can be used to quantify corruption provides an opportunity not only to describe the phenomenon more precisely, but also to predict it. However, the problem of prediction is not rigorously addressed within the vast academic literature about corruption, and we know little of the aspects that may help predict these acts in a precise manner (Colonnelli, Gallego & Prem, 2019). For example, the statistical models used in diverse studies (Olken & Pande, 2012) lack the capacity of prediction due to the scope of the data used, the spatial and temporal scales of analysis, the fields of operation, and the high dependency on correlations, making nearly impossible to establish causalities for different events (Riccardi & Sarno, 2014).…”

“…As it is, conclusions and proposals derived from such models should be considered reservedly within their respective contexts. Nevertheless, with current computing capabilities and trends towards the digitalization of information in many public sectors, the ideal and promising approaches to predict corruption risk seem to be those supported by livemonitoring systems powered by artificial intelligence (Fazekas & Kocsis, 2017;López-Iturriaga & Sanz, 2018;Colonnelli et al, 2019). Such systems may provide a wide range of real-time preventive capabilities (Byers, 2017;Altshuler & Pentland, 2018).…”

“…Noteworthy, the identification of latent criminal groups (Campedelli, Cruickshank & Carley, 2019) and the effective dismantling of their organizational structure (Wandelt, Sun, Feng, Zanin & Havlin, 2018) are relevant and non-trivial subjects in criminal investigations and law enforcement, since empirical evidence has shown that the dismantling process might potentially make these criminal organizations stronger (Duijn, Kashirin & Sloot, 2014). In addition, when it comes to fighting corruption the goal is clear: one not only is looking to describe it post factum, but to predict it (Rumi, Deng & Salim, 2018;Alves, Ribeiro & Rodrigues, 2018;López-Iturriaga & Sanz, 2018;Colonnelli et al, 2019;Colliri & Zhao, 2019).…”

Inefficient Procurement in Times of Pandemia

“…In line with previous research (Gallego et al, 2019;Colonnelli et al, 2020), we construct an objective measure based on observable characteristics and a machine learning approach. This measure allows us to the disentangle the effects of corruption on public procurement in the midst of an emergency.…”

Corruption and Firms

“…Since the CGU construct these measures only starting in 2006, we compute the number of irregularities in the 2003-2005 audits using the manually constructed data inColonnelli et al (2020b). 41 Specifically, following the approach ofColonnelli et al (2020a), we estimate a LASSO model using a set of 147 municipality characteristics and a ten-fold cross-validation to find the parameter that best fits the data. We then standardize the predicted probability by subtracting its mean and dividing by its standard deviation.42 Using numbers from the empirical distribution of the standardized share in the data, we can compute that municipalities with predicted levels of detected corruption in the bottom 10th percentile of the distribution experience a negative effect on total firms of 4.3%.…”