3
$\begingroup$

Let's split each period into $n$ intervals. There's a continuum $u$ of unemployed and $v$ of vacancies. During each interval, there is a total of $X/n$ job offers. That means that each unemployed gets a job offer with probability $X/(nu)$ (assuming that $n$ is eventually so small that the chance of multiple job offers to the same person will go to zero).

Single individual Number of job offers - for a single individual - in the interval is distributed $Binomial(k, n, \frac{X}{un})$. Letting $n\to\infty$, the distribution in the continuous-time analog converges to $Poisson(k, \frac{X}{u})$.

I'm interested in the probability of $x$ individuals getting at least one job interval during a whole interval. That is $(1 - Binomial(0, n, \frac{X}{nu}))^x$:

$$ (1 - (1 - \frac{X}{un})^n)^x $$

If I'm not mistaken, the continuous-time analog is

$$ (1 - e^\frac{-X}{u})^x$$

Is that correct? The combination of exponential and power function is making me quite uncomfortable.

$\endgroup$

2 Answers 2

2
$\begingroup$

An alternative approximating approach you could use as as check might be to say there are $X$ job offers in total and $u$ unemployed.

So the probability that an individual does not get a particular job offer is $\left(1-\dfrac{1}{u}\right)$ and so the probability the individual does not get any job offer is $\left(1-\dfrac{1}{u}\right)^X$ which is $\left(\left(1-\dfrac{1}{u}\right)^u\right)^{x/u} \approx e^{-X/u}$ much as one might expect from a Poisson distribution. You have an error in the sign of $e^{-X/u}$.

That makes the probability the individual does get a job offer about $1-e^{-X/u}$. It makes the expected number of individuals receiving job offers about $u(1-e^{-X/u})$.

The probability that $y$ individuals get job offers (rather than $x$ as you have already used $X$) is then roughly the binomial $\displaystyle {y \choose u}\left(1-e^{-X/u}\right)^y e^{-(u-y)X/u}$ and or if you wanted a Poisson approximation to this then about $\dfrac{e^{-u\left(1-e^{-X/u}\right)}u^y\left(1-e^{-X/u}\right)^y}{y!}$ and you could apply further approximations to this but it would not get much tidier

$\endgroup$
2
  • $\begingroup$ If $u$ is a continuum, isn't the probability of a particular individual getting the job $0$? I thought about this case first, but to get the math correctly, one would - I thought - have to have time in intervals $n$, and some measure of agents $m$, and look at the double limit when $n\to0$ and $m\to\infty$. $\endgroup$
    – FooBar
    Sep 13, 2016 at 6:11
  • $\begingroup$ Your splitting into $n$ intervals now works to get you to $(1 - e^\frac{-X}{u})$ for the probability a particular individual gets at least one job offer. Currently your problem is turning that into a probability that a given number of individuals get offers, and for that you have the binomial distribution rather than just putting it to the power of $x$ (or $y$ in my version). $\endgroup$
    – Henry
    Sep 13, 2016 at 7:29
2
$\begingroup$

Yes, it is correct. You can (for instance) write a Taylor expansion: \begin{align*} [1-(1-\frac{x}{un})^n]^x & = [1-e^{n ln(1-\frac{x}{un})}]^x \\ & = [1-e^{n (-\frac{x}{un} + o(\frac{1}{n}))}]^x \\ & = [1-e^{-\frac{x}{u} + o(1)}]^x \\ & \sim [1-e^{-\frac{x}{u}}]^x \text{ when } n \rightarrow +\infty \end{align*}

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.