Quantitative Evaluation Techniques for Regional Policies

In recent years, the vision of what the essential factors for growth are and therefore the role of local policies has drastically changed. The importance of aspects such as human capital, innovation, agglomeration and institutions coupled with the diversified impacts of globalization, have drawn attention to the often-neglected role of space for growth and growth policies (Barca et al. 2012). Moreover, the presence of a wide and persistent inequality in income and joblessness among local areas, regions and countries, exacerbated by the Great Recession, has suggested a more important role for spatially targeted policies. Austin et al. (2018) indicate that place-based policies should be considered in this framework, because “social problems are increasingly linked to a lack of jobs rather than a lack of income” and “subsidizing job creation may be easier at the place level than at the person level”. Barca et al. (2012) argue that “space matters and shapes the potential for development not only of territories, but, through externalities, of the individuals who live in them.” Therefore, the place-based approach is more appropriate than a space-neutral sectoral approach if the geographical context matters, in terms of social, cultural, and institutional characteristics. These considerations have led to a new spread of place-based policies, often accompanied by skepticism with respect to their results from a significant group of economists and politicians (see, for instance, Glaeser and Gottlieb 2008). Indeed, “a fundamental concern is that spatially targeted policies may simply shift economic activity from one locality to another, with little impact on the aggregate level of output” (Kline and Moretti 2014). It is therefore not surprising that in recent years there has been a particular effort in the development of techniques capable of evaluating the effectiveness of territorial policies. In this survey of place-based policy evaluation techniques, we have chosen to consider only methodologies and studies based on the counterfactual approach. The reason is that we are convinced that to identify the effects of a policy we need a causal model, and the counterfactual approach is the most widely used and convincing approach in this field. The counterfactual approach, typical of program evaluation literature, attempts to compare what actually happened with what would have happened in the absence of the treatment. As each unit can be exposed or not exposed to the treatment (see Holland 1986), the researcher is bound to compare treated units with distinct untreated units. This approach derives from the potential outcomes framework (see Rubin 1974) where pairs of outcomes are defined for the same unit given different levels of exposure to the treatment, with the researcher only observing the potential outcome corresponding to the level of treatment received. Models are developed for the pair of potential outcomes rather than solely for the observed outcome. The potential outcomes framework has a number of advantages over a framework based directly on realized outcomes: i) it allows one to define causal effects before specifying the assignment mechanism, and without making functional form or distributional assumptions; ii) it forces the researcher to think about scenarios under which each outcome could be observed, that is, to consider the kinds of experiments that could reveal the causal effects; iii) it allows formulation of probabilistic assumptions in terms of potentially observable variables, rather than in terms of unobserved components; iv) it separates the modeling of the potential outcomes from that of the assignment mechanism. Of particular importance in Rubin’s approach is the relationship between treatment assignment and the potential outcomes (Imbens and Wooldridge 2009). The simplest case for analysis is random assignment of the treatment, which ensures that there are no systematic differences between the treatment and control groups before treatment assignment. This implies that any observed differences in outcomes following the treatment can then be attributed to the treatment itself, rather than to selection bias. Therefore, it is straightforward to obtain estimators for the average effect of the treatment. Randomized experiments have been used in some areas in economics but hardly ever in regional economics. This is why in this survey we will focus on observational studies.