Serving the Quantitative Finance Community

 
User avatar
Alan
Topic Author
Posts: 2958
Joined: December 19th, 2001, 4:01 am
Location: California
Contact:

How much volatility is due to "trading"?

January 10th, 2003, 4:48 pm

There is a school of economists who believe that the stock market has "too much" volatility.I am not an economist, but this seems silly to me. Regardless of the "fundamentals", it'speople who price things and they can be fickle if they want, change their risk attitudes,horizons, whatever. But I am curious as to forum members opinions about how muchvolatility we see in the stock market is due to "trading" vs. fundamentals. By fundamentalvolatility, I just mean the volatility that *would occur* because a company makes a widget and themarket for that has its ups and downs and so does their earnings. And by *would occur*,I guess I mean that there isn't much trading, so say the market is only open for 15 min.once a week to set a new price and that's it. Then, any volatility beyond that is due totrading.
 
User avatar
reza
Posts: 6
Joined: August 30th, 2001, 3:40 pm

How much volatility is due to "trading"?

January 10th, 2003, 5:20 pm

Alan, this is a very philosophical questionI am not an economist by any means.but how would one separate the fundamentals from trading?trading by definition is the meeting of supply and demand driven by a desire for profitso which part is impact of news? and which news?hard to say ...this is the kind of question that Aaron would like
 
User avatar
Alan
Topic Author
Posts: 2958
Joined: December 19th, 2001, 4:01 am
Location: California
Contact:

How much volatility is due to "trading"?

January 10th, 2003, 5:54 pm

I don't think it's so philosophical. To quantify it a little more, supposethe NYSE actually does decide to open just once a week for 15 min. Now we can measure price returns weekly and there will be a weeklystandard deviation of returns sigma. So I am asking for opinions about theratio R of the weekly sigma in this new arrangement vs. the weeklysigma you have now. Do you think R = 0.1? or R = 0.98? or R > 1?. Would it depend upon the stock?. How about for the SPX? There is some old information about this kind of thing from an oldstudy when the NYSE had to close once a week, maybe it wasweds, to clear trades. Weekend closings give less information sincethey are not business days. And overnight gives less for the same reason.
 
User avatar
JabairuStork
Posts: 0
Joined: February 27th, 2002, 12:45 pm

How much volatility is due to "trading"?

January 10th, 2003, 6:08 pm

Well, the trivial answer is that ALL volatility comes from trading, because if nobody traded prices wouldn't move (unless you count matrix pricing, but that is for sissies.)However, there is definitely an element of price movement which is autocorrelated, and this autocorrelation has different properties when measured over different time scales. So if you had a system where all trades get filled only once a week, that should destroy any autocorrelation which exists only at frequencies higher than weekly. I guess the question is whether the autocorrelation that exists at lower frequencies is independent of the higher frequency part. We should be able to decompose the overall time series of prices into waveforms of various frequencies that are independent, but theoretically I'm not convinced this is entirely valid and practically there is nowhere near enough data to do this reliably.In the UST market you have a situation where you have auctions at specified dates, which set OTR prices. The only difference is that the securities trade on the secondary market in between auctions. If you carefully adjust for the OTR liquidity premium and the pull-to-par effect, you might be able to make a valid comparison of variance in auction prices vs. variance in secondary market prices for govvies.Any thoughts?
 
User avatar
Johnny
Posts: 0
Joined: October 18th, 2001, 3:26 pm

How much volatility is due to "trading"?

January 10th, 2003, 6:29 pm

"There is a school of economists who believe that the stock market has "too much" volatility. I am not an economist, but this seems silly to me."When an economist says that there is "too much" volatility he means that markets are more volatile than his theories suggest. The economist wants to improve his theory of volatility. I don't understand why that should be any more silly than trying to have a theory of what price something should be. To quantify it a little more, suppose the NYSE actually does decide to open just once a week for 15 min. Now we can measure price returns weekly and there will be a weekly standard deviation of returns sigma. So I am asking for opinions about the ratio R of the weekly sigma in this new arrangement"Here you have made a dramatic change to the market micro-structure. How much of the change in volatility will be due to the change in market micro-structure? How much to the trading? And how much to changes in trader's risk aversion? (e.g. job risk of not getting an order filled ... or filled at VWAP or whatever). It's not obvious at all that this is something that is easy to measure. However, to someone trying to design an efficient securities exchange these are important questions.Suppose that you want to price a call option on an illiquid share. How do you do this without having a theory of volatility? Well, obviously you guess, because no-one has a theory this good. But in principle this would be a theory worth having. Presumably it would be some kind of general equilibrium theory that could solve for the market price of volatility at the same time as the market price of the security. But I don't see that this is either "silly" or easy to quantify.
 
User avatar
Alan
Topic Author
Posts: 2958
Joined: December 19th, 2001, 4:01 am
Location: California
Contact:

How much volatility is due to "trading"?

January 10th, 2003, 6:47 pm

QuoteOriginally posted by: JabairuStorkAny thoughts?Good observation about auto-correlation. However, suppose we write the weekly (log) return as the sum of 5 dailies. Now suppose there are two different "worlds", both hypothetical, so maybereza is right, this is philosphy :-)In world I, let's say each day's return is drawn independently from some stationary distribution,with variance sig^2. (not necessarily normal). So there is zero auto-correlation. Inworld I, trading hours are like today's real world.In world II, trading opens up once a week as my example, but it's populatedby exactly the same people as in world I.Then, in principle, couldn't we have 5 sig^s (world I)either much greater or much less than the weekly return variance in world II?So I guess I'm suggesting its not just an auto-correlation issue, although that is clearly part of theanswer to my question.
 
User avatar
Anthis
Posts: 7
Joined: October 22nd, 2001, 10:06 am

How much volatility is due to "trading"?

January 10th, 2003, 7:03 pm

Prices (quote mid-price) move because of older flow differentials. Volume just represents opinion and expectations heterogeneity. This is a mix of true (and possibly assymetric) information and noise. The greater the heterogeneity the greater the volume. Order flow differentials are mainly due to information which can be separated into payoff information and non payoff information. The rest is due to liability reasons in the framework of an Asset Liability Management system. For example a mutual fund manager must buy (sell) if he has net inflows (outflows) of cash, that is when investors tend to buy (sell) shares of his fund, pushing consequently the prices of his asset holdings up (down). If people had homogeneous expectations about an asset then no trade would take place after the price had matched those expectations unless there is a liquidity/liability motivation for trade.Anthis
 
User avatar
Alan
Topic Author
Posts: 2958
Joined: December 19th, 2001, 4:01 am
Location: California
Contact:

How much volatility is due to "trading"?

January 10th, 2003, 7:45 pm

<i>When an economist says that there is "too much" volatility he means that markets are more volatile than his theories suggest. The economist wants to improve his theory of volatility. I don't understand why that should be any more silly than trying to have a theory of what price something should be.</i> The "silly" part, to me, is a theory that leads to a conclusion that observed levels of market volatility are grossly too high or"irrational". The reason I say this is that what little I know of rational equilibrium-type pricing always postulates preferenceparameters like risk-aversion and time-preference. In general, and please correct me if I am worng, I believe there's nothing in this general theory to prevent these from changing dramatically, let's say every day. Given that, then "rational" traders can create whatever volatility they want . If a particular theory leads to predicting dramatically lower volatility than what is observed, I would guess it's because it doesn't allow for changing preferences. So the silly part seems to be neglecting changing preferences and then concluding "irrationality". Maybea better adjective would be "wrong-way conclusion". The effort itself of trying to construct a volatility theory is certainly not silly. <i>Here you have made a dramatic change to the market micro-structure. How much of the change in volatility will be due to the change in market micro-structure? How much to the trading? </i>By "trading", I really just meant this change in the market micro-structure. So my question is really about the amount of volatility that peoplebelieve is due to that. Thanks for helping me clarify that ambiguity.
Last edited by Alan on January 9th, 2003, 11:00 pm, edited 1 time in total.
 
User avatar
Anthis
Posts: 7
Joined: October 22nd, 2001, 10:06 am

How much volatility is due to "trading"?

January 10th, 2003, 11:42 pm

-----------The "silly" part, to me, is a theory that leads to a conclusion that observed levels of market volatility are grossly too high or "irrational". The reason I say this is that what little I know of rational equilibrium-type pricing always postulates preference parameters like risk-aversion and time-preference.----------Think that there is a metric of return per unit of risk like Sharpe ratio. Certainly it cant be steady over time. If this metric is currently below its historical long run average then it would make me conclude that volatility is high in a given market. Preferences and utility functions change if at all only slowly and gradualy and the main reason are changes in a person's income level due to the law of diminishing marginal utility.Furthermore relaxing the rationality assumption you can easily see that markets do overreact to news creating short term trends and reversals.Eventually if NYSE was about to change in one weekly trading session I I expect that the only difference you may observe are huge price jumps during the weekly market opening (when the price formation takes place) relevant to the previous week’s close priceAnthis
 
User avatar
Alan
Topic Author
Posts: 2958
Joined: December 19th, 2001, 4:01 am
Location: California
Contact:

How much volatility is due to "trading"?

January 11th, 2003, 12:09 am

QuoteOriginally posted by: Anthis<brPreferences and utility functions change if at all only slowly and gradualy and the main reason are changes in a person's income level due to the law of diminishing marginal utility.Eventually if NYSE was about to change in one weekly trading session I I expect that the only difference you may observe are huge price jumps during the weekly market opening (when the price formation takes place) relevant to the previous week’s close priceAnthisThanks, Anthis, you are the first to offer an opinion, which is that the volatility wouldn't much change.Could you elaborate on why you think preferences and utility functions only change slowly? What abouttrader's income? :-) Seriously, my sense is that people change their "appetite for risk" quite a lot, but it's mostly gut feel, so I am curious to know why you think the opposite.
 
User avatar
David
Posts: 2
Joined: September 13th, 2001, 4:05 pm

How much volatility is due to "trading"?

January 11th, 2003, 12:46 am

In a world where trading is permitted only once a week for 15-min, the information set I (increasing sequence of sigma field) will undoubtedly be very big, but the volatility might stabilized and even decreased slightly at some point. In general, the difference between actual volatility and realized volatility should be correlated, at least in the theory, but in reality...? The market needs certain time to find the equilibrium (or to assimilate) when the fundamentals are out at the opening or during the trading session. So in the 15-min world, the intra-day price will fluctuate wildly until the equilibrium would be found (if at all) in the same day. However, nothing guarantees that the weekly volatility will be high as well, but it will be possible to observe a sizeable deviation between actual volatility and the realized volatility.
 
User avatar
Anthis
Posts: 7
Joined: October 22nd, 2001, 10:06 am

How much volatility is due to "trading"?

January 11th, 2003, 12:46 am

Trader's income? Clarify please. I guess you dont mean the "rogue trader" paradigm. If yes it is an issue of principal-agency theory
 
User avatar
Anthis
Posts: 7
Joined: October 22nd, 2001, 10:06 am

How much volatility is due to "trading"?

January 11th, 2003, 12:50 am

DavidI assume that the opening call auction can find the new equilibrium price rapidly.
 
User avatar
Alan
Topic Author
Posts: 2958
Joined: December 19th, 2001, 4:01 am
Location: California
Contact:

How much volatility is due to "trading"?

January 11th, 2003, 1:10 am

QuoteOriginally posted by: AnthisTrader's income? Clarify please. I guess you dont mean the "rogue trader" paradigm. If yes it is an issue of principal-agency theoryFirst, David, the intraday volatility for the 15 min doesn't matter to me. I just picked 15 min. as a reasonabletime to let the market clear, as Anthis said. Any time interval would do, as long as its long enough to seta new price, but short enough so that there is not much "business news" that occurs duringthe interval.Anthis, I don't know what the rogue trader paradigm is. I was making a joke. So what is it?In trying to look up some references on google, I see a more standard way to phrase myquestion for the economists. From somebody's homepage:------------------------------------------------------------------------------------------------------------------- "endogenous volatility" refers to economic fluctuations that arise purely from interactionsof market participants, rather than from exogenous shocks that might impact the economy.-------------------------------------------------------------------------------------------------------------------So I am asking for opinions on the relative proportions of endogenous to exogenous volatilityin the total volatility that we observe.
 
User avatar
David
Posts: 2
Joined: September 13th, 2001, 4:05 pm

How much volatility is due to "trading"?

January 11th, 2003, 1:12 am

Anthis,I disagree. Take the non-farm payrolls for example, it takes usually 10min-30min to find the new equilibrium.