In the long term the Bitcoin price could be the most important factor affecting difficulty levels, as with higher Bitcoin prices greater resources are used for mining Bitcoins, pushing up hash rates and then difficulty levels.
In the medium term those ASIC bulk chip sales from Avalon will be the main consideration. The extra estimated 84 TH/s, at around $2.8 million in chip value, could have an enormous impact on mining output and difficulty levels thereafter, expected at end July 2013. ASICminer and products from other producers will also feature.
In calculating return on investment miners have to consider firstly the Bitcoin price and secondly the technology improvements allowing a faster hashrate per KW/h. It is hard to predict either future Bitcoin prices or technology improvements though in the medium term technology advancements are mostly visible. In the longer term an adaption of Moore's law could apply to mining, if you consider the upgrade path we have seen for mining from Central processor, Graphics Processor, FPGA then ASIC etc.
I have plotted the following graph for a two year period comparing difficulty and Bitcoin price.
For most uses the graphical plot is quite satisfactory to judge the correlation by eye.
Whether there is evidence of the causal relationship for a econometrics statistician to prove a hypothesis on past data is relatively easy to establish by simply plugging the data into a advanced financial econometrics software program which gives you the probability of relationships and other measures such as the error variables, lag and noise. I was about to do this when I discovered that while I have accurate Bitcoin price data, the data for past daily difficulty levels I had gathered, while fine for examining graphical correlation, that it may not be 100% accurate for other uses due to time Unix period between measures not being exactly 24 hours i,e, +{Time}/60/60/24+25569.
I sourced the data here:
blockexplorer.com/q/nethashIf anyone can point me to a source of 100% accurate daily historic data for difficulty levels from day one of Bitcoin to the present it would be very helpful indeed.