Accounting for the bias against the life-cycle hypothesis in survey data: An example for Russia