Smithian State Formation

We examine the relationship between land inequality and state capacity, measured with tax collection. Contrary to traditional models that emphasize the redistributive role of taxation, we focus on its role in providing public goods that help develop markets, increasing productivity (e.g. enforcement of property rights, coordination, roads, electricity). We build a simple model of public good provision where landowners decide whether or not to comply with property taxes taking into account: (i) that government expenditure increases property values, and (ii) their expectation of the punishment when evading taxes. We validate the model empirically, using data from Colombian municipalities between 1923 and 1960, in two ways. First, we use detailed land values' data from cadastres available for a subsample of municipalities and the model's structure to predict tax revenues and compare them with actual revenues. Second, we provide empirical evidence supporting the model's main prediction: land concentration is positively correlated with tax revenues per capita, a relationship that is robust to controlling for potential sources of omitted variable bias. Two additional empirical findings further support the mechanisms underlying the model: the positive relationship is stronger in places with greater potential gains from market development, and land concentration reduces the average fiscal cost of collecting one peso of revenue.*****ABSTRACT: We examine the relationship between land inequality and state capacity, measured with tax collection. Contrary to traditional models that emphasize the redistributive role of taxation, we focus on its role in providing public goods that help develop markets, increasing productivity (e.g. enforcement of property rights, coordination, roads, electricity). We build a simple model of public good provision where landowners decide whether or not to comply with property taxes taking into account: (i) that government expenditure increases property values, and (ii) their expectation of the punishment when evading taxes. We validate the model empirically, using data from Colombian municipalities between 1923 and 1960, in two ways. First, we use detailed land values' data from cadastres available for a subsample of municipalities and the model's structure to predict tax revenues and compare them with actual revenues. Second, we provide empirical evidence supporting the model's main prediction: land concentration is positively correlated with tax revenues per capita, a relationship that is robust to controlling for potential sources of omitted variable bias. Two additional empirical findings further support the mechanisms underlying the model: the positive relationship is stronger in places with greater potential gains from market development, and land concentration reduces the average fiscal cost of collecting one peso of revenue.