Although it should, but do we really take into account their area while calculating efficiency because in that particular area no light is actually absorbed.
When calculating efficiency limits of solar cells, contact shadowing is not taken into account since this is an avoidable loss mechanism. Contact shadowing can be eliminated by putting all contacts on the back side, as in an interdigitated back contact cell.
Shadowing losses caused by the frontal electric grid are both an avoidable efficiency loss AND a meaningful terms when optimizing your cell design for maximum efficiency.
This is especially true when you try to set the trade-off between the reduced Joule heating losses and increased optical losses due to increasing the area/number of your frontal grid pattern.
Efficiencies are usually calculated for the ENTIRE device area, so front-side metallisation leads to loss, and as Omar said, the metal grid is optimised for the best balance of optical loss and resistive loss.
However, sometimes parts of the grid (especially the busbar which connects all the other metal lines, and which can even be physically outside the cell area) are not included. This is why in efficiency tables such as Green 2015 PiP (http://onlinelibrary.wiley.com/doi/10.1002/pip.2573/full), a comment referring to the way the area is defined (total area, designated illumination area) is added.
The standard requires that you consider all the area of the cell. By doing so, you can compare results between cells regardless of their finger dimensions and thinner grid wires.
Another point is that the reflection from these metal grids is actually mostly downwards towards the cell. I quote from a paper:
The effective shading of metal‐plated solar‐cell fingers after cell encapsulation is substantially less than would be expected from simple geometric calculations. The reason for this is that much of the light reflected from the fingers is either directed downwards toward the silicon or is totally internally reflected at the air/glass interface. Experimental data is presented to support calculations of this effect. Calculations of the impact of this effect on solar‐cell performance are also included.