Basics in AC circuits
Since we are using AC circuits, we have inputs which can be described as:
$$I_{in} = A\sin(\omega t + \phi)$$
And:
$$V_{in} = B\sin(\omega t + \phi)$$
Note that all sinusoid input signals can be written like this! (If the signal is not a pure sinusoid, we can use Fourier decomposition, but this is beyond the scope of this post)
The differential equations for inductors and capacitors are:
$$V_L = L\frac{dI_L}{dt}$$
$$I_C = C\frac{dV_C}{dt}$$
If we would take a single inductor or capacitor and hook it up to a periodic input signal, then $I_L = I_{in}$ and $V_C = V_{in}$. We can now calculate $I_L$ and $V_L$:
$$V_L = AL\omega\cos(\omega t + \phi)$$
$$I_C = BC\omega\cos(\omega t + \phi)$$
Here the terms $L\omega$ and $C\omega$ can be identified as the reactances $X_L$ and $X_C$ (note $\omega = 2 \pi f$).
As you can see, current leads voltage in the capacitor, and voltage leads current in the inductor. This can be remembered via "CIVIL".
We can now go to phasor notation, as the only thing important in those circuits is the phase. This is keep track of the phases. As we will see, phases are very important in AC circuits, especially when dealing with power.
Phasor notation
We can convert a periodic signal, like $V_L$ to phasor notation, which is:
$$V_L = AL\omega\cos(\omega t + \phi) = \Re(L\omega e^{i\phi}e^{i\omega t})$$
You can check this for yourself. In phasor notation we drop some terms; the signal in phasor notation is:
$$V_L = AL\omega e^{i\phi}$$
Complex power is defined as:
$S = UI^* = P + iQ = \Re(UI^*) + i\Im(UI^*)$
Note the complex conjugate! Here P is the active power and Q the reactive power (to be elaborated below).
In a resistor, where the current and voltage are in-phase, we can calculate the power by integrating over one period of the signal. This power is called active power, and yields the familiar (?) $\frac{1}{\sqrt{2}}$ term.
In inductors and capacitors, it is a different story, however. Since we are integrating a sine and cosine over one period, where the sine and cosine are orthogonal functions, this yields zero power! This can be interpreted in the case of the inductor as first storing power into the magnetic field and then retrieving this power again. This doesn't look like a problem, right?
In power networks, this is a problem! This power actually has to flow through the load lines, and then flows back again! We want to use $active$ power, not this "weird" power, which is called $reactive$ power.
The answer
Now to come back to the original question:
If we would express our phasors in terms of the voltages (so a current phasor in a capacitor would have a 90 degrees positive phase with respect to the voltage phasor, and a current phasor in an inductor would have a 90 degrees negative phase with respect to the voltage phasor), then we have that: (assuming phi is in the range (-180, 180] degrees
If $\phi = 0$ degrees or $\phi = 180$ degrees we have $active$ power.
If $\phi < 0$ degrees we have inductive reactive power.
If $\phi > 0$ degrees we have capacitive reactive power.
In other words, the reactances of capacitors are $iX_c$ and those of inductors have $-iX_l$, where the i's deal with the positive/negative 90 degrees phase shift of the phasor.