[Следующее, возможно, кажется немного техническим из-за использования уравнений, но оно основано главным образом на стрелочных диаграммах, чтобы обеспечить интуицию, которая требует только очень элементарного понимания OLS - так что не отталкивайтесь.]
xiyiβ
yi=α+βxi+↖corrϵi↗
This might have happened because we forgot to include an important variable that also correlates with xi. This problem is known as omitted variable bias and then your βˆ will not give you the causal effect (see here for the details). This is a case when you would want to use an instrument because only then can you find the true causal effect.
An instrument is a new variable zi which is uncorrelated with ϵi, but that correlates well with xi and which only influences yi through xi - so our instrument is what is called "exogenous". It's like in this chart here:
zi→xi↑ϵi→↗yi
So how do we use this new variable?
Maybe you remember the ANOVA type idea behind regression where you split the total variation of a dependent variable into an explained and an unexplained component. For example, if you regress your xi on the instrument,
xitotal variation=a+πziexplained variation+ηiunexplained variation
then you know that the explained variation here is exogenous to our original equation because it depends on the exogenous variable zi only. So in this sense, we split our xi up into a part that we can claim is certainly exogenous (that's the part that depends on zi) and some unexplained part ηi that keeps all the bad variation which correlates with ϵi. Now we take the exogenous part of this regression, call it xiˆ,
xi=a+πzigood variation=xˆi+ηibad variation
and put this into our original regression:
yi=α+βxˆi+ϵi
Now since xˆi is not correlated anymore with ϵi (remember, we "filtered out" this part from xi and left it in ηi), we can consistently estimate our β because the instrument has helped us to break the correlation between the explanatory variably and the error. This was one way how you can apply instrumental variables. This method is actually called 2-stage least squares, where our regression of xi on zi is called the "first stage" and the last equation here is called the "second stage".
In terms of our original picture (I leave out the ϵi to not make a mess but remember that it is there!), instead of taking the direct but flawed route between xi to yi we took an intermediate step via xˆi
zi→xi↗→xˆi↓yi
Thanks to this slight diversion of our road to the causal effect we were able to consistently estimate β by using the instrument. The cost of this diversion is that instrumental variables models are generally less precise, meaning that they tend to have larger standard errors.
How do we find instruments?
That's not an easy question because you need to make a good case as to why your zi would not be correlated with ϵi - this cannot be tested formally because the true error is unobserved. The main challenge is therefore to come up with something that can be plausibly seen as exogenous such as natural disasters, policy changes, or sometimes you can even run a randomized experiment. The other answers had some very good examples for this so I won't repeat this part.