Dornbusch Model


The relative stickiness of prices and wages means that the exchange rate will overshoot in response to changes in monetary policy.


  • Like portfolio-balance models Dornbusch assumes, home and foreign goods are imperfect substitutes, that good markets adjust slowly relative to asset markets so that domestic prices and real output are fixed in the S-R, domestic money is only held in the home country, and the domestic country is small relative to world asset and goods marekts so that it faces a fixed foreign interest rate and price level. 
  • Key differing assumption: domestic and foreign interest-earning assets are perfect substitutes. I.e. i=i*+pi. Thus do not need equations for B & F because their earlier role was one of determining i & i* separately.
  • pi = theta(e-bar-e) i.e. partial adjustment to exchange rate toward an equilibrium level e-bar.
  • Also, wealth effects have no role in determining exchange rates so can focus on money market equilibrium for S-R exchange rate determination
  • Demand for money=M/P=Y^phi*exp^-(lamba*i) which in logarithms is m-p=phi*y-lambda*i where m, p & y are the logarithms of money, price and real income


Key equation:

  • e=e-bar-1/(lambda*theta)*[p-m+phi*y-lambda*(i*)]
2016-04-09 15.02.37.jpg

For a given price level the exchange rate always adjusts instantaneously to clear the money market and thus always on the AA line. But long-run equilibrium requires being on the 45-degree line which has the exchange rate moving proportionally with the price level to maintain eP*/P in line with PPP (P* is assumed constant).


In the S-R prices and wages are largely inflexible leaving only exchange rates to change in response to monetary policy. Thus the exchange rate will temporarily overshoot to Q before returning to R. 


The model explains the surge in the US dollar from 1980 to 1984 well above its expected PPP value after the Fed tightened monetary policy unexpectedly in 1979-80 before falling again in 1985.


However if rigidities in nominal wages and prices are to blame then its not clear that deviations should persist for so long - Friedman argued they should only take 12-24 months.