Understanding arousal/performance relationships
Back up to Arousal control
Four models of the effects of arousal - physical or mental - on performance have similar implications. They're certainly oversimplified, and none is 'quantitative'. I've put them here for two reasons. First, they show how understanding of arousal and performance is developing, and also increasngly recognising the complexity of the problem. Second, the pictorial representation can be useful in talking about the practical effects of arousal; for example, "The Zone" is a description arising from the second of these pictures.
 Drive theory
Drive theory simply says "More pressure gives greater performance". True to a point... for example, with no pressure at all, you don't perform at all. Or get out of bed, for that matter
The "Inverted U" hypothesis (Note 1) starts like drive theory, but predicts a point where increasing arousal decreases performance. That sounds more familiar - and it certainly happens that way in endurance events.
The hysteresis model recognises that in many cases, following a ‘crash’ one has to reduce arousal substantially before performance starts to recover much. That's a pretty good description for physiological arousal, but maybe not so good for 'anxiety'.
 Multidimensional models
Multidimensional models recognise many components to ‘arousal’; a physiological component (heart rate etc.), a ‘cognitive’ component (nervousness, fear of failure or consequences of failure), and further factors which reduce or increase the problem. The one above (Note 2) suggests that as anxiety ("cognitive anxiety") increases, 'falling off a cliff' is more likely, but confidence acts to reduce the effect.
SOME arousal is essential for good performance These two add up to something important. Somewhere, there is a Zone where performance is best - an optimum arousal level, or optimum arousal point. Finding that Zone is the main aim of arousal control.
Too much arousal is bad for performance
Recovering from a ‘crash’ is possible but often slow  
Many different factors affect 'arousal'  

Note 1. I'm reminded that this was first described by Yerkes and Dodson in the 1920's!

Note 2. This particular picture is based on a model suggested by Fazey and Hardy, in 1988, which attempts to account for the observation that in some circumstances, performance follows the inverted U scheme and degrades smoothly with increasing arousal, but in others, the decline is sudden and steep. They suggested that the transition from smooth to catastrophic behaviour was related to cognitive anxiety. The model is sometimes referred to as 'catastrophe theory' (though I suspect that's more because the phrase was trendy then than because it's got much to do with the mathematical concept of catastrophe theory ;o) ) and, like everything that fits in one paragraph, I've oversimplified the implications.