A Globally Stable Price Adjustment Process (Q3979447)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 19932
Language Label Description Also known as
default for all languages
No label defined
    English
    A Globally Stable Price Adjustment Process
    scientific article; zbMATH DE number 19932

      Statements

      A Globally Stable Price Adjustment Process (English)
      0 references
      0 references
      0 references
      26 June 1992
      0 references
      price adjustment
      0 references
      Walras' tâtonnement
      0 references
      global Newton method
      0 references
      It is well-known that Walras' tâtonnement may fail to converge to an equilibrium unless an additional assumption, such as the gross substitutes condition, is made. The purpose of this paper is to present a differential equation which, under quite general conditions, converges to an equilibrium unless the initial price vector belongs to a set of measure zero in the domain. We consider an economy with \(\ell\) goods. Let \(z: R^ \ell_ +\backslash \{0\}\to R^ \ell\) be the excess demand function. Using homogeneity of degree zero, Walras' law, and a boundary condition, we may use \(T=\{\hat p=(p_ 1,\ldots,p_{\ell-1})\in R_ +^{\ell-1}\mid\sum^{\ell-1}_{i=1}p^ 2_ i\leq 1\}\) as the domain, and the excess demand function may be written as \(\hat z: T\to R^{\ell- 1}\). Walras' tâtonnement and Smale's global Newton method [see \textit{S. Smale}, J. Math. Econ. 3, 107-120 (1976; Zbl 0354.90018)]\ have the forms \((d\hat p/dt)=\hat z(\hat p)\) and \(D\hat z(\hat p)(d\hat p/dt)=- \kappa\hat z(\hat p)\), respectively, where \(D\hat z(\hat p)\) denotes the Jacobian matrix of \(\hat z\) at \(\hat p\), \(\det D\hat z(\hat p)\) is its determinant, and \(\kappa\) is a real number such that \(\text{sign} \kappa=\text{sign}\det(-D\hat z(\hat p))\). Our process is a weighted average of the above two processes where the weights depend on the norm of the excess demand and the distance between \(\hat p\) and the initial price vector \(\hat p^ 0\). Namely, our process is \[ ({D\hat z(\hat p) \over \parallel \hat z(\hat p)\parallel}-{I \over \parallel \hat p-\hat p^ 0 \parallel}) {d\hat p \over dt}=-\lambda \hat z(\hat p),\leqno(1) \] where \(I\) is the \((\ell-1)\times (\ell-1)\) identity matrix, \(\parallel \cdot \parallel\) denotes the Euclidean norm, and \(\lambda\) is a real number such that \(\text{sign} \lambda=\text{sign} \det(I/\parallel \hat p- \hat p^ 0\parallel-D\hat z(\hat p)/\parallel \hat z(\hat p)\parallel)\). Although \(d\hat p/dt\) is not directly defined at \(\hat p=\hat p^ 0\), it can be appropriately defined by taking a limit. Our process always converges to an equilibrium unless the initial price vector \(\hat p^ 0\) belongs to a set of measure zero in \(T\). Note that, at \(\hat p^ 0\), the process is Walras' tâtonnement, and as it approaches an equilibrium it becomes Newton's method.
      0 references

      Identifiers