In this paper we present a generic framework for the asymptotic performance analysis of subspace-based parameter estimation schemes. It is based on earlier results on an explicit first-order expansion of the estimation error in the signal subspace obtained via an SVD of the noisy observation matrix. We extend these results in a number of aspects. Firstly, we derive an explicit first-order expansion of the Higher- Order SVD (HOSVD)-based subspace estimate. Secondly, we show how to obtain explicit first-order expansions of the estimation error of ESPRIT-type algorithms and provide the expressions for matrix-based and tensor-based Standard ESPRIT and Unitary ESPRIT. Thirdly, we derive closed-form expressions for the mean square error (MSE) and show that they only depend on the second-order moments of the noise. Hence, we only need the noise to be zero mean and possess finite second order moments. Fourthly, we investigate the effect of using Structured Least Squares (SLS) to solve the overdetermined shift invariance equations in ESPRIT and provide an explicit first-order expansion as well as a closed-form MSE expression. Finally, we simplify the MSE for the special case of a single source and compute the asymptotic efficiency of the investigated ESPRIT-type algorithms in compact closed-form expressions which only depend on the array size and the effective SNR. Our results are more general than existing results on the performance analysis of ESPRIT-type algorithms since (a) we do not need any assumptions about the noise except for the mean to be zero and the second-order moments to be finite (in contrast to earlier results that require Gaussianity or second-order circular symmetry); (b) our results are asymptotic in the effective SNR, i.e., we do not require the number of samples to be large; (c) we present a framework that incorporates various ESPRIT-type algorithms in one unified manner.