Imagine a NYSE with 20% less Volatility!
Imagine the changes it would cause for our economy. I got some feedback from a macroeconomist friend on how a reduced volatility in the NYSE equities market would affect our economy.
I believe that this could happen if we helped to save Social Security by starting a public mutual fund, US Mutual, like what I've described before.
The idea is that US Mutual would be a low overhead mutual fund that would use 28 or so weeks of data and conservative median log-return(=Ln(1+return)) predictions and their median standard deviations to rank stocks with an index. It would then use the same index to determine its long holdings in the top 20% of common stocks(weighted for total valuation) and short holdings in the bottom 5% of stocks.
It would exhibit economies of scale, targeting a 6% control of NYSE with 4% spread out over the top 20% and 2% focused on the bottom 5%. The selling short on the unstable underpeformers is what would reduce the overall volatility of the NYSE.
Some have brought up the issue of market power. I agree that the initial weeks of starting US Mutual would probably require expert assistance and maybe a little bit of persuasion, as the top companies would need to be reminded of how it is in their interest to maintain the stability of their stock value and the good will of the US gov't by not let the selling price for their stock rise up too dramatically.
It would be more of a matter of negotiation, a win-win situation for the companies and US Mutual that would ensure that the US public will benefit from these preferred stocks.
I think that it bears corroboration with historical data and simulations. I've already done some preliminary work that shows that median log-return regressions predict better than mean or median return regressions or mean log-return regressions. These stats are critical to the algorithm in that they help to keep the holdings patterns more stable so the levels of buying and selling of stocks on a weekly basis would be limited and US Mutual would use its considerable market power more as a force for stability than instability in the NYSE.
dlw
I've had a small change in the idea. I've thought a little bit more about the predictions that would be used to determine holdings.
I think the holdings would also be based on predicted steady-state log-returns and the steady state median standard deviations(presuming that an AR(1) regression would be significant for weekly log-return regressions). Ie., let's say we regressed ln(return) on its lagged value and an intercept. We would find the steady state value of ln(return) by assuming that the ln(return) and its lagged value were equal in value. We could then with some simple algebra solve for the predicted steady-state values. These predicted steady-state values will not change much on a weekly basis, unlike predictions that use the last weekly values.
So since we want to value stability more with US Mutual, the predicted values used to determine US Mutual's holdings would weight the predicted steady-state values twice as much as the predicted values using the previous week's values. This would limit further the impact of the relative volatility on the changes in the holdings of US Mutual. It would help to guarantee that US Mutual's market power would be more of a force for stability in the NYSE...
dlw
A blog-activist dedicated to the reduction of the faith-based political acrimony in the United States of America so as to make our political system more democratic and just and to improve our witness to the rest of the world.
July 17, 2006 3:24 PM | Reply | Permalink
I've had a small change in the idea. I've thought a little bit more about the predictions that would be used to determine holdings.
I think the holdings would also be based on predicted steady-state log-returns and the steady state median standard deviations(presuming that an AR(1) regression would be significant for weekly log-return regressions). Ie., let's say we regressed ln(return) on its lagged value and an intercept. We would find the steady state value of ln(return) by assuming that the ln(return) and its lagged value were equal in value. We could then with some simple algebra solve for the predicted steady-state values. These predicted steady-state values will not change much on a weekly basis, unlike predictions that use the last weekly values.
So since we want to value stability more with US Mutual, the predicted values used to determine US Mutual's holdings would weight the predicted steady-state values twice as much as the predicted values using the previous week's values. This would limit further the impact of the relative volatility on the changes in the holdings of US Mutual. It would help to guarantee that US Mutual's market power would be more of a force for stability in the NYSE...
dlw
A blog-activist dedicated to the reduction of the faith-based political acrimony in the United States of America so as to make our political system more democratic and just and to improve our witness to the rest of the world.
July 17, 2006 3:24 PM | Reply | Permalink