A "snapshot" of Robert Jarrow's talk on Detecting Asset Price Bubbles in Real Time
Detecting Bubbles
Three ways to detect in real time:
1. modeling the fundamental value (not practical),
2. modeling the process S (successful use is outlined below),
and
3. modeling derivatives (still untested).
Detecting Bubbles - Joint Hypo with {S}
Empirically test this by:
1. estimating s(x) on observed asset price levels x (we use a
local time estimator),
2. extraplating to rest of domain x 2 [0,¥) (we use best t to
di¤erent parametric functions, e.g. s0xa or s0xa lnb jxj),
3. checking if the integral is nite for the estimated parametric
function.
Can independently test for t of asset price evolution.
Partially avoids curse of joint hypothesis.
Conclusions
_ It is possible to test for price bubbles in real time!
_ Illustrated method based on stochastic asset process.
_ Method using derivatives is still untested.
(Full presentation to be posted when available.)
Replies