In general, for CMOS chips, power consumption is made up of two main parts - static and dynamic. Static power is independent of clock frequency, while dynamic power is consumed when transistors switch and as such is dependent on clock frequency. Dynamic power consumption of a given transistor is computed as \$P = C v^2 \alpha f\$, where \$C\$ is the load capacitance, \$v\$ is the supply voltage, \$\alpha\$ is the activity factor, and \$f\$ is the frequency. \$C\$ is defined by the layout (wire capacitance and gate capacitance of driven transistors), \$\alpha\$ is representative of how often the transistor switches, and \$v\$ must be high enough to ensure reliable operation at a given \$f\$.
If you overclock by changing the clock frequency only, then the dynamic power consumption will change linearly while the static power consumption stays constant. If the static power consumption is zero, then the change will be directly proportional. But the static power consumption is likely to be rather high, especially with the leaky transistors used in modern processes, and on a modern CPU not every component runs on the core clock anyway, so the change in power consumption will likely be less.
However, a 30% overclock is quite high and likely won't be possible without also increasing the core voltage. In that case, not only will the static power consumption increase, the \$v^2\$ term also comes in to play and you can end up with a very large increase in power consumption.