Starting from the initial consumption bundle, first the substitution effect is calculated, then the income effect. Why not the other way around?

My guess would be, because with increasing utility the shape of the utility level curve might change, the substitute effect would differ not just proportionally, while there is no such problem with income effect. However, I don't see why the initial utility level curve is preferred to the new one.

One could argue, that the initial utility curve is unrepresentative, because it doesn't represent the current situation.

  • $\begingroup$ Hi! It seems to me that in case of Slutsky (or Hicks) decomposition, what you actually calculate are three consumption bundles. The pre price change choice $(A)$, the post price change choice $(C)$, and the income compensated choice given the new price $(B)$. Once you have these three bundles, you can freely calculate the income $(x_1^C - x_1^B)$ and substitution $(x_1^B - x_1^A)$ effects in the order you want? Do I misunderstand your question? $\endgroup$
    – Giskard
    Commented Dec 17, 2021 at 17:01
  • $\begingroup$ Hey! I was talking about taking point B as the choice with the old price and the income adjusted (reduced if prices rise) so as to be on the new indifference curve. But as you put it this way, income compensated sounds to have a better meaning $\endgroup$ Commented Dec 18, 2021 at 7:43
  • $\begingroup$ Okay, but is anything stopping you from calculating $x_1^B - x_1^A$ before calculating $x_1^C - x_1^B$? $\endgroup$
    – Giskard
    Commented Dec 18, 2021 at 9:28

1 Answer 1


That's actually just a convention. While these two methods to partition the total effect lead to different results for any discrete price change, they coincide in the limit of an infinitesimal price change.

  • $\begingroup$ Are you writing about Slutsky decomposition here? If yes, what two methods? I seem to have lost the thread. $\endgroup$
    – Giskard
    Commented Dec 17, 2021 at 15:25
  • 1
    $\begingroup$ @Giskard, I was actually referring to this problem economics.stackexchange.com/questions/417/… $\endgroup$
    – VARulle
    Commented Dec 20, 2021 at 8:43

Your Answer

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

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