Most evidence of hyperbolic discounting is based on violations of either stationarity or time consistency as observed in choice experiments. These choice reversals may however also result from time-varying discount rates. Hyperbolic discounting is a plausible explanation for choice reversals only if violations of stationarity and time consistency overlap. Our field experiment examines the extent to which this is the case. At different points in time, the same participants allocated a future gift over sooner-smaller and later-larger rewards with varying front-end delays. We find that most violations of time consistency do not coincide with violations of stationarity. This is surprisingly similar to what an earlier experiment on stationarity, time invariance and time consistency finds using a different design among a different type of participants (Halevy, Econometrica , 2015). Random noise in decision-making alone does not explain this finding, given that we find a significant association between changes in household wealth and violations of stationarity and time consistency. We conclude that when incomes fluctuate, one can only identify hyperbolic discounting by eliciting violations of both stationarity and time consistency through a longitudinal design for the same subject pool.