(John Kemp is a Reuters market analyst. The views expressed are his own)
* Chart 1: tmsnrt.rs/2jnAqnM
* Chart 2: tmsnrt.rs/2jnAB2q
* Chart 3: tmsnrt.rs/2jnvuiO
By John Kemp
LONDON, Jan 18 (Reuters) - "We are entering a period of much more volatility in the (oil) market", the head of the International Energy Agency warned on Sunday.
"The name of the game is volatility" Fatih Birol said in an interview with Reuters Television, repeating similar comments he made at the end of 2016.
Birol is among many analysts who have warned that the outlook for oil prices is choppier and more volatile in 2017 but the concept needs to be employed carefully.
Volatility has always been one of the defining characteristics of oil prices and is apparent at all time scales from the very short term (seconds, minutes and hours) to long periods (months, years and decades).
And volatility is coupled with a pronounced cyclical behaviour in which periods of rising and falling prices alternate ("Volatility and cyclicality in oil prices - will this time be different?" Reuters, Jan. 13).
But the concept is notoriously slippery and how traders talk about it does not necessarily correspond with the way in which the concept is understood by producers and consumers.
Volatility has a precise definition in finance where it is a statistical measure of the dispersion (standard deviation) of price movements.
Volatility is normally measured over a short interval (most often daily price changes), evaluated over a short period of time (usually over 20 or 30 days) and expressed at an annualised rate.
Volatility as the term is employed by traders is a good measurement of how much prices jump around over short time scales.
But in ordinary language the term is used to refer to big price swings such as the near-quadrupling in oil prices between 2002 and 2008 and the more-than-halving of prices between 2013 and 2016.
Micro-volatility as understood by option traders is not the same as the macro-volatility as understood by oil producers and consumers.
If oil prices rose by 1 percent per day for 100 days, micro volatility would be zero but macro volatility would be very high.
Micro volatility is concerned with the smoothness with which prices move while macro volatility focuses more on the overall scale of the cyclical swings.
Oil prices can jump around a lot from day to day (in which case micro volatility is high) while remaining trendless (macro volatility is low).
Or prices can move only a small amount each day (low micro volatility) but exhibit large and sustained rising or falling trends (high macro volatility).
In a further wrinkle, price volatility can be measured in terms of either dollars per barrel or percentage price changes.
Percentage changes tend to be higher when oil prices are low because the same dollar move is a greater proportion of the base price.
If oil prices jump around by $1 per day percentage volatility is much higher when the base price is $30 than if it is $100.
Oil refiners tend to be more worried about dollar changes because their throughput margins are based on dollars per barrel.
Crude option traders tend to focus on percentage price moves because these are the inputs into their pricing models.
Researching cotton prices, mathematician Benoit Mandelbrot discovered there were far more very large daily price moves than would be expected if changes followed a normal or Gaussian distribution.
Mandelbrot also discovered price changes alternated between periods of low and high volatility ("The (mis)behaviour of markets", Mandelbrot, 2004).
The typical amount of volatility is not constant but is itself volatile, with markets experiencing abrupt phase shifts from "mild" to "wild" and back again.
Mandelbrot was writing about micro volatility but his observations about the volatility of volatility also apply at macro scales.
Oil prices exhibit both regular up-and-down cycles from one month or one year to the next, as well as huge price spikes and crashes (tmsnrt.rs/2jnAqnM).
Regular cycles pass almost unremarked while spikes and crashes catch the popular imagination as booms, busts and super-cycles.
The recent record shows that oil prices were not especially volatile during 2015 and 2016 in the micro-sense of the term despite the price slump (tmsnrt.rs/2jnAB2q).
Percentage volatility has been higher than during the period from 2011 to 2014 but this seems to have been mostly due to the fall in dollar prices (tmsnrt.rs/2jnvuiO).
Micro volatility has generally remained moderate with the exception of surges following production cuts announced by the Organization of the Petroleum Exporting Countries in September and November 2016.
Macro volatility has been higher, with prices doubling between January 2016 and January 2017, after slumping by more than two-thirds during the previous 18 months.
But when Birol and other analysts talk about prices being more volatile in 2017, there is no reason to think they will be more volatile in the micro sense as employed by option traders.
Instead what they seem to be saying is that prices will fluctuate or cycle around a level rather than trend upwards as they did for much of 2016.
In this view, high prices will stimulate more shale production but also risk curbing consumption growth, while low prices will curb shale drilling and risk provoking action from OPEC.
Many analysts believe price cycles will be shorter and shallower owing to the faster response times of shale producers, though this theory remains unproven and is controversial.
If it is true, price cycles in 2017-2018 are unlikely to resemble the boom-bust cycles of 2002-2009 and 2009-2016.
Instead, the new normal in 2017-2018 may resemble the period between mid-2011 and mid-2014, when prices regularly cycled in a $25 range between about $95 and $120.
Shorter price cycles may not be much of a problem for oil producers and consumers provided prices fluctuate within a reasonably stable and predictable band.
Consumers and producers can cope with a lot of volatility at the micro and even intermediate scales provided there is not too much volatility at macro scale.
Stability in a band of $50-$60 or $45-65 per barrel is good enough for most producers and consumers to plan their businesses.
Fluctuations only become problematic if the range starts to stretch much below $45 or above $65 in which case they are likely to become more destabilising.
Editing by David Evans