Il giorno giovedì 17 giugno, alle ore 16:30, Kent Osband (Chief Risk Officer, SBlocks, e già World Bank, IMF, Rand Corporation) terrà sulla piattaforma Zoom il seminario:
What Can Capital Markets Teach Us About Learning?
What can capital markets teach us about learning? Not much, says orthodox finance, since the market as a whole already knows. Not much, says behavioral finance, since markets behave irrationally despite strong incentives not to. With those caveats, let us imagine a capital market dominated by artificially intelligent virtual robots (“bots”), who (i) know that trends might change in ways they cannot fully foresee, (ii) rationally digest all available information to learn all they can, and (iii) aim to maximize their long-term wealth by gambling on their predictions.
Both capital markets turn out to have an uncanny resemblance to real life-markets. Prices are excessively volatile compared to dividends, and exhibit GARCH-like behavior with volatile surges in volatility.
Disagreements are rife, with trading far too frequent to ascribe to risk management alone. Equity risk premia are high despite modest risk aversion. Debt markets persistently warn too little too late about
unsustainable sovereign debt burdens.
This seminar will provide a non-technical overview of these findings and draw out some broader implications. These include:
• Human minds work like capital markets, with attention a rough analogue to capital.
• Rapid learning generates turbulence, with disagreements flaring before they subside.
• Respect for disagreement facilitates learning, as there are huge benefits to doubt.
• Capital markets are generally poor long-term forecasters and great error-correctors.
• Banking regulation should require far larger capital buffers than currently maintained.
• Bond markets are unlikely to give much advance warning of monetary disaster.
Modalità di partecipazione e allegati
I partecipanti potranno partecipare cliccando sull’apposito link Zoom
Tutto il materiale disponibile per il seminario (slides, working paper, locandina) è scaricabile presso questo indirizzo
Il working paper è disponibile anche a quest’altro indirizzo