Second, we observe a negative impact of the density of contracted specialists on the one of non-contract physicians, indicating relatively strong competition forces among these physicians.
Therefore, we estimate the densities of (i) non-contract specialists and (ii) non-contract general practitioners as a function of the densities of contract general practitioners and specialists and the remaining density of non-contract physicians, along with other control variables such as the availability of private and public inpatient facilities within a district.
We estimate the local density of one type of non-contract physicians (PGPs or PSPs) as a function of the densities of contract physicians (CGPs and CSPs) and the remaining density of non-contract physicians, among other factors such as a region's hospital facilities, educational level or aggregate income.
Basically, the HT-estimator starts with the consistent FE-estimates, takes the within residuals of this regression, say e ˆ it, and, in a second step, regresses e ˆ it on X i using the time-variant exogenous variables as instruments (in our case, the densities of contract physicians).
We also observe a negative relationship between the densities of non-contract and contract general practitioners and the ones of non-contract and contract specialists, indicating competitive forces between the private and the public sector of the outpatient health care provision in Austria.
We investigate the density of non-contract (private) physicians in a two-tiered health care system, i.e., one with co-existing public and private health care providers.
Further, we assume that the right-hand-side densities of both contract physicians and public hospital beds are exogenous, which might be justified by the fact that they are determined by the physician capacity and hospital plans.
First, we find a positive association between the densities of non-contract general practitioners and non-contract specialists, indicating a relatively strong referral effect between these groups of physicians.
In our case, this approach is useful to address the potential endogeneity of the right-hand-side densities of non-contract physicians and private hospital beds.
