1: \begin{abstract}

2: This study considers a model of the

3: income distribution of agents whose

4: pairwise interaction is asymmetric and

5: price-invariant. Asymmetric

6: transactions are typical for

7: chain-trading groups who arrange their

8: business such that commodities move

9: from senior to junior partners and

10: money moves in the opposite direction.

11: The price-invariance of transactions

12: means that the probability of a

13: pairwise interaction is a function of

14: the ratio of incomes, which is

15: independent of the  price scale or

16: absolute income level. These two

17: features characterize  the hierarchical

18: model. The income distribution in this

19: class of models is a well-defined

20: double-Pareto function, which possesses

21: Pareto tails for the upper and lower

22: incomes. For  gross and net upper

23: incomes, the model predicts definite

24: values of the Pareto exponents, $a_{\rm

25: gross}$ and $a_{\rm net}$, which are

26: stable with respect to quantitative

27: variation of the pair-interaction. The

28: Pareto exponents are also stable with

29: respect to the choice of a demand

30: function within two classes of

31: status-dependent behavior of agents:

32: linear demand ($a_{\rm gross}=1$,

33: $a_{\rm net}=2$) and unlimited slowly

34: varying demand  ($a_{\rm gross}=a_{\rm

35: net}=1$). For the sigmoidal demand that

36: describes limited returns, $a_{\rm

37: gross}=a_{\rm net}=1+\alpha$, with some

38: $\alpha>0$ satisfying a transcendental

39: equation. The low-income distribution

40: may be singular or vanishing in the

41: neighborhood of the minimal income; in

42: any case,  it is $L_1$-integrable and

43: its Pareto exponent is given

44: explicitly.

45: 

46: The theory used in the present study is based on a simple

47: balance equation and new results from

48: multiplicative Markov chains and

49: exponential moments of random

50: geometric progressions.

51: \end{abstract}

52: