One of the well known paradoxes in macroeconomics is that estimates of risk aversion from experimental micro data do not match the ones estimated from macro data.
I know there was an important paper that pointed this out first but I forgot the name of author(s) and title. Does anyone know what paper it is?
In an extremely influential paper, Mehra and Prescott (the reference is below) showed that in order to explain the premium of stocks over treasury bonds in a parametric general equilibrium model, one needs to assume levels of risk aversion that are a magnitude higher than what one usually assumes. However, the difference is not so much with microeconomics than what is reasonable for other problems. The difference between microeconomics and macroeconomics is not that clear though. It should also be noted that risk aversion in laboratory experiments is much higher than what appears reasonable. For the tiny sums involved, one would expect subjects to behave as if they were risk-neutral. They do not.
Mehra, Rajnish, and Edward C. Prescott. "The equity premium: A puzzle." Journal of Monetary Economics 15.2 (1985): 145-161.
The literature was initially quite pessimistic about the possibility to consistently estimate risk-aversion parameters consistently (and more generally all models' parameters), as highlighted by
Carroll, Christopher Dixon, 2001, "Death to the Log-Linearized Consumption Euler Equation! (And Very Poor Health to the Second-Order Approximation)," The B.E. Journal of Macroeconomics, 1(1).
However, provided the time horizon is long enough, there is some hope to estimate the parameters consistently from micro-data, as discussed by:
Attanasio, Orazio, P. and Hamish Low, 2004, "Estimating Euler equations," Review of Economic Dynamics, 7, 405–435.
Regarding the macroeconomic literature, the issue of consistent estimation cannot be investigated at all with macro data, as aggregation over individuals is only possible under strict conditions. For a study illustrating how "the size of the bias increases with the level of aggregation", see for instance:
Cutanda, A., J.M. Labeaga, and J.A. Sanchis-Llopis, 2020, "Aggregation biases in empirical Euler consumption equations: evidence from Spanish data," Empirical Economics, 58, 957–977.
Well, this does not identify which is precisely the original paper showing the inconsistency between the micro and macro estimates, but rather tend to give support to the thesis according to which macro estimates are inconsistent, and micro estimates can be consistent for the true parameters, under some reasonable conditions.
