Solving a complex system of equations with summation

64 Views Asked by At

Could anyone please give any idea on how the following system of equations has to be solved with respect to vector $\hat W$ (which, essentially, contains all the unknowns in the system).

\begin{cases} \hat L_{i,j} = (\sigma - 1)\sum_{k=1}^9\nu_{k,j}\hat p_{k,j} \space - \space \sigma\hat p_{i,j},\\ \hat p_{k,j}=\tau_{k,j} + \hat w_{k},\\ \sum_{j=1}^9s_{i,j}\hat L_{i,j} = \beta\hat w_i \end{cases}

The model describes change in employment $\hat L_{i,j}$ in region i because of a change in supplies to region j.
$\sigma, \beta $ are exogenous constants. $\nu$ is a matrix of constants ($\nu_{k,j}$ is a share of region's j income spent on goods from region k), $s$ is a matrix of constants ($s_{i,j}$ is a share of region's i production directed to region j), $\tau$ is a matrix of constants as well ($\tau_{k,j}$ represents costs related to trade between region k and j). $p$'s are prices of goods and $w$'s are wages in corresponding regions. There are 9 regions in total.

I can't figure out the way to find a general solution to the system (preferably, in matrix form, which I can't derive for this system either). Writing them all down doesn't really seem rational since there're like 81+81+9 equations.

If there's no general solution, how do I aproach it then to find vector $\hat W$? Any ideas? Thanks in advance!

1

There are 1 best solutions below

0
On BEST ANSWER

$$ \sum_{j=1}^9s_{i,j}\left((\sigma - 1)\sum_{k=1}^9\nu_{k,j}\left(\tau_{k,j} + \hat w_{k}\right) \space - \space \sigma( \tau_{i,j} + \hat w_{i})\right) = \beta\hat w_i $$

$$ (\sigma - 1)\sum_{j=1}^9s_{i,j}\left(\sum_{k=1}^9\nu_{k,j}\tau_{k,j}\right)+(\sigma - 1)\sum_{j=1}^9s_{i,j}\left(\sum_{k=1}^9\nu_{k,j}\hat w_{k}\right)-\sigma\sum_{j=1}^9s_{i,j}\tau_{i,j}-\sigma\sum_{j=1}^9s_{i,j} \hat w_{i} = \beta\hat w_i $$

$$ \beta\hat w_i+\sigma\sum_{j=1}^9s_{i,j} \hat w_{i}+(\sigma - 1)\sum_{j=1}^9s_{i,j}\left(\sum_{k=1}^9\nu_{k,j}\hat w_{k}\right) = (1-\sigma)\sum_{j=1}^9s_{i,j}\left(\sum_{k=1}^9\nu_{k,j}\tau_{k,j}\right)-\sigma\sum_{j=1}^9s_{i,j}\tau_{i,j} $$

$$\left(\beta+\sigma\sum_{j=1}^9s_{i,j}\right)\hat w_i+(\sigma-1)\sum_{j=1}^9s_{i,j}\left(\sum_{k=1}^9\nu_{k,j}\hat w_{k}\right) = (1-\sigma)\sum_{j=1}^9s_{i,j}\left(\sum_{k=1}^9\nu_{k,j}\tau_{k,j}\right)-\sigma\sum_{j=1}^9s_{i,j}\tau_{i,j}$$

Now define $S$, $V$, $T$, $\hat{\mathbf w}$ as matrices/vectors where $[S]_{i,j}=s_{i,j}$, $[V]_{k,j}=\nu_{k,j}$, $[T]_{i,j}=\tau_{i,j}$, $[\hat{\mathbf w}]_{i}=\hat w_{i}$, as well as $\mathbf 1=[1,1,1,\ldots,1]^t$ (a column vector of 9 ones) and I think we have (you should check the details)

$$\left(\beta+\sigma \mathbf 1^tS^t+(\sigma-1)SV^t\right)\hat{\mathbf w} = (1-\sigma)S\mathbf 1^t(V\circ T)-\sigma(S\circ T)\mathbf 1$$

where $A\circ B$ is the Hadamard product (element-wise multiplication). This is a linear system but I don't think you can expect to find a general solution, at least it's a simple expression to give to a numerical solver.