We have an externality here because the private cost to me differs from the cost my action imposes on society as a whole. If the benefit to me exceeds the cost to society of waiting until a less congested time to travel, I should travel anyway. But what if the benefit to me exceeds my private cost (I'd rather travel on Friday, even though the trip will be longer and more aggravating than if I waited), but is short of the social cost? I travel, and society is harmed. On balance the net benefit to me is less than the costs that my trip imposes on others.
The trick for society is to get me to take account of these costs. One way to do so is to impose a tax on drivers during peak periods. If the tax is set correctly (a non-trivial requirement), society can get the right amount of congestion (It is not zero). Such taxes force me to take into account the effects of my actions on others, since I pay if I go on Friday. People who absolutely, positively have to use the road on Friday night do, and pay the tax. Others, who can travel more flexibly, do so, and save the tax payments.
Such taxes are called Pigouvian taxes, after an English economist,
Arthur Cecil Pigou, who spent much of his career on questions of
externalities. 
Arthur Cecil Pigou If one agrees that
externaities are pervasive in a market economy, and that Pigouvian
taxes are a good solution to the problems they pose, one has the
making of a case for government intervention to correct market
failures. Indeed, while we do not ordinarily impose congestion taxes
on prime time drivers, intervention to internalize externalities is
pervasive in our market system.
The process is at work at the Ohio State parking garages. OSU faculty pay a substantial sum for parking in OSU garages. This means that to get faculty members to work at OSU, their compensation will need to be high enough relative to alternatives to cover the costs of parking. Why not just charge zero for parking, but lower salaries? This strategem would lower the cost of faculty to the university, for the parking fees are paid in after-tax income. But despite the tax-man's bite, parking fees continue to be charged to make drivers internalize the costs they impose on others when they use a parking space.
We are coming closer to legal disputes here. When I drive a car recklessly, I do so with the understanding that I may injure myself and/or my property. But I could also injure you and yours. Do I take the latter into account? Not if I don't have to pay in the event of an accident. We can interpret the compensation to my victims I expect to pay as a Pigouvian tax that attempts to generate the right amount of accidents by making me consider all of the costs of my actions.
Notice that I may still choose to drive recklessly. I may think the benefit to be is worth the costs, even when I bear all of the costs of my actions. A polluter may be willing to bear all costs of the pollution pumped into the environment. But in such cases, its better to put up with some accidents, or some pollution, than to pay the costs of not engaging in the activities in question.
In a regime of zero transactions costs, lawyers would perish.Suppose that my house is located next to your fragrance factory. Your plant puts out noxious fumes, noxious to me at least, though perhaps not objectionable to others. Suppose that the cost to me of experiencing these fumes is $100 per year. Suppose also that the fumes can be cleaned up for only $75 per year. What is the correct tax? Should it be $100 dollars annually? If we impose a tax of this magnitude, what happens? The fragrance firm installs the abatement equipment, the tax is not paid, I get clean air, and society is out $75 in real resources, the resources used for the abatement equipment. This is a $25 net improvement over doing nothing. Is this the correct result?George J. Stigler
Memoirs of an Unregulated Economist
page 76.
Not necessarily. I am $100 worse off if I have to experience the stench. Whose fault is that? Is it the fault of the factory? Is it my fault? Suppose I could move away from the plant, selling my home to an insensitive individual not bothered by the plant, for a net cost of $25. What should happen? I'm the lowest cost avoider. The best response to the externality is for me to move away. That results in a real resource cost to society of only $25. But if the Pigouvian tax is imposed, I have no incentive to move. The smell is gone, but at a cost of $50.
Now suppose that both the plant owner and I are reasonable people. What do you think will happen? Should the plant owner spend $75 to obtain a benefit worth $25 to me? Suppose that the plant owner offers me $50 to put up with the smell. Will I take the deal? Sure. I'll move away, incurring $25 in real resource costs, and pocketing the extra $25. The plant owner is also ahead $25—he does not now need to install the pollution control equipment (unless the government is intransigent on the point). Of course, the joint $50 benefit from negotiations only remains if the costs in real resources of actually cutting a deal between us are zero.
Suppose that the numbers has been the same, but the government had not intervened at all, so that my choices are to stay and take the smell, to move, or to offer a payment to the factory owner to stop the smell. What happens? I move. I offer no more thatn $25 to the factory owner, an offer that is rejected. The result is the same as above, when the factory owner was responsible, except that the $50 transfer payment to me is not made. That is, real resource use is unaltered, but the distribution of income is different.
We have the makings of a theorem, which we will call the Coase
Ronald Coase
Theorem, in honor of its inventor, Ronald Coase. If (and this is a
very big If) we live in a world where people can make
bargains costlessly, externalities cease to exist. And, even more
surprisingly, in this world where transactions are costless, neither
the assignment of legal liability for damages nor legal ownership of
resources has an effect on the utilization of real resources.