Learning from experience and trading volume

We build a model where trading allows inexperienced agents to discern useful information sources. Upon losing money by trading on invalid information sources, investors learn from their experience and switch to alternative sources. Such activity leads to initial expected losses but later profitable trades. Trading activity is found to be increasing in the mass of such agents. Volume is greatest in firms with uncertain cash flows. Further, a greater number of information sources implies greater volume. This is consistent with the explosive growth in volume accompanying the growth of the internet, which presumably increases the number of heterogeneous information sources.