*I show that if predictable factors drive stock returns, they admit a vector autoregressive (VAR) representation. I establish an equivalence between the predictability of financial markets and an increase in uncertainty and systemic risk on the markets. Modularity is introduced as a measure of uncertainty and systemic risk. Unlike the existing measures of uncertainty, it is easier to calculate for any financial market and produces market-specific information. I establish that an increase in uncertainty or systemic risk leads to a collapse of the (trust) linkages between firms listed on financial markets. This result is related to the theory of reputation and fragility of financial markets. I prove that a financial crisis is not due to an increase in the linkages between financial firms but to an increase in uncertainty on the market, unlike what financial market regulators assume. I present the implications for the regulation of financial markets. I also show that the modularity of non-financial firms and mid-cap firms can warn one year in advance about the explosion of financial bubbles. These results are valid across Global, US, European and French markets over 2005-2020. However, the estimation of modularity requires estimating a high-dimensional VAR model. For this reason, I introduce a general framework for estimation and inference in high-dimensional linear regression models with stationary data.

**This paper analyzes how the effect of low general practitioners’ density on quality of primary care is mediated by enrollment in integrated practices.
We use two independent cross-sectional surveys (2019-2020) of 1,209 French private practice general practitioners (GPs) representative of those practicing in France (except Mayotte) in terms of gender, age, workload and GP density of the practice location.
We assumed that low GP density influences their wellbeing as well as their use of e-health tools and ultimately the (1) quantity of care approximated by a latent variable comprising GPs’ declared practices regarding patient care management and (2) quality of care approximated by a latent variable comprising frequencies of vaccine recommendations. Moreover, we consider an enrollment in an integrated practice (so-called multi-professional group practice) as an additional mediator between GP density and the quality of care.
Our results suggest that the relation between low GP density and quantity of care delivered within their practice in the absence of potential mediators was negative and significant; there was no significant effect on vaccine recommendations. In the presence of mediators, low GP density was significantly and negatively associated with GPs’ wellbeing, that was consecutively associated with inferior patient care management. Conversely, there was a positive and significant correlation between low GP density and an enrollment in an integrated practice, mediated by the use of e-health tools which was in turn associated with greater involvement in vaccine recommendations.