Lecture 3: The Rural-Urban Wage Gap: The Labor Turnover Model

• The basic question here is why urban employer is not reducing the wage rate even if there is
unemployment. There are many explanations for this. One of the reasons is that frequent labour
turnover is costly for the employer, particularly in the manufacturing sector.
• The whole assembly line gets disturbed till a new worker matches the rhythm of other workers.
Also in industry, workers usually work with machines and efficient operation of the machinery
achieve only through learning by doing or experience.
We assume the following: Joseph Stiglitz
(1974). “Wage
1. We assume that turnover of labour does not impose any costly adjustment process in rural
determination and
agriculture.
unemployment in
2. It is also postulated that firms paying higher wage rate faces lower turnover rate.
LDCs: the labor
turnover model”.
• The above points provides an intuitive answer to the question we posed at the beginning, and this In: Quarterly Journal
also basis of a model by Stiglitz (1974)9, which endogenous the rural-urban wage gap. of Economics 88(2),
pp.194–227
• Here we consider a simplified version of the Stiglitz’s labour turnover model.
• The annual turnover or quite rate of labour faced by a firm depends on many factors, of
which Stiglitz consider three.
1. The wage paid by the firm vis-´ a-vis the average industrial wage,
2. The wage paid by the firm vis-´ a-vis the rural wage, and
3. The urban unemployment rate

• The above three factors signals the general labour market conditions, and if any one of
them deteriorates workers would be hesitant to quit a secure job in search of a better one.
• For the time being we assume that there is only one urban firm, later we
consider more than one urban firm and interaction among firms in the
context of labour turnover.
• While urban-rural wage ratio and urban unemployment are factors that
determine the quit rate, in this model these two are causally related. Hence
we consider only one of them as determinant of the quit rate, q
𝑤
q=q , 𝑄′ ≤ 0
𝑤𝑅

• where 𝑤 is the urban wage and 𝑤𝑅 is the rural wage, which is equal to the
marginal product of labour.
• The production function in the urban sector is
𝑋𝑀 = 𝑓𝑀 𝐿𝑀 , 𝑓𝑀′ > 0, 𝑓𝑀′′ < 0
𝑊
• The firm faces two cost: wage cost (W𝐿𝑀 ) and the labour turnover cost 𝑇𝑞 𝐿𝑀 ,
𝑊𝑅

where T is the cost incurred by the firm every time a worker quits and a new person is
employed. Here it is assumed quit worker is immediately replaced and the firm faces no
labour shortage.
• The firm chooses w and 𝐿𝑀 so as to maximise profit, π, given by
𝑤
π = 𝑓𝑀 𝐿𝑀 -w𝐿𝑀 − 𝑇𝑞 𝐿𝑀
𝑤𝑅

𝑤
which may be re-written as π = 𝑓𝑀 𝐿𝑀 − 𝑤 + 𝑇𝑞 𝑤𝑅
𝐿𝑀

𝑤
As obvious from above, the firm first chooses w to minimise 𝑤 + 𝑇𝑞 and then chooses 𝐿𝑀 to maximise π.
𝑤𝑅

𝑇 𝑤
It is clear from the following first-order conditions: 1+ 𝑤 𝑞 ′ =0
𝑅 𝑤𝑅

𝑤
𝑓 ′ 𝐿𝑀 = 𝑤 + 𝑇𝑞 𝑤𝑅

The above two equations can be solved to get w and 𝐿𝑀 as functions of 𝑊𝑅 and T, however since T remains unchanged, it
may be suppressed.
w= w 𝒘𝑹 ………………(1)
𝑳𝑴 = 𝑳𝑴 𝒘𝑹 ……………….(2)
• Thus, once the rural wage rate (𝑊𝑅 ) is known, we know the levels of urban
employment and urban wage.
• To keep the analysis simple, assume that marginal product of the labour in rural
sector is constant ‘m’
• This implies a production function of the following type 𝑋𝑅 = m𝐿𝑅 .
• Since wage equals the marginal product of labour, 𝑊𝑅 is constant. Therefore by the
above equations, w and 𝐿𝑀 are given and the only variable to be explained is the
rural employment 𝐿𝑅 ..
Following Harris-Todaro, we assume that workers migrate to a sector
which offer higher expected wage and that urban employment rate
represents the probability of finding an urban job (in the rural sector
jobs are assured). Then the migration equilibrium condition is,

𝐿𝑀
𝑊𝑅 = 𝑊
𝐿 − 𝐿𝑅
where L denotes the total labour available in the economy.
• The equilibrium condition given in the previous slide is meaningful if and
only if w ≥ 𝑤𝑅 .
If 𝑤𝑅 > w then every one would join the rural labour force, that is L = 𝐿𝑅 .
• It means that the full statement of the migration equilibrium condition is as
follows: if w ≥ 𝑤𝑅 , then the above equilibrium condition is valid, and if 𝑤𝑅 >
w, then 𝐿𝑅 = L.
• For the time being assume that w ≥ 𝑤𝑅 . And the case 𝑤𝑅 > w will be
considered later.
• Having solved (1) and (2) for w and 𝐿𝑀 , we can solve the migration equilibrium
given above for 𝐿𝑅 .
• Let [𝑊 ∗ , 𝐿∗𝑀 , 𝐿∗𝑅 ] be the solution vector obtained from the above procedure.
• Since 𝑊 ∗ ≥ 𝑊𝑅 , 𝐿∗𝑀 + 𝐿∗𝑅 ≤ 𝐿. Therefore [𝑊 ∗ , 𝐿∗𝑀 , 𝐿∗𝑅 ] ] is the equilibrium vector in
the model.
• Before exploring the model critically, we explain it diagrammatically. This would
help us to see the kind assumptions that quit function and production function need
to satisfy to ensure the existence and uniqueness of a solution.
• The figure presents the 𝑇𝑞 as a function of w, for a given
𝑊𝑅 .
𝑤
• It is assumed that if is very high, greater than θ, say,
𝑤𝑅

𝑤
then q′( ) = 0.
𝑤𝑅

• In other words q(θ) is the “natural” quite rate determined

by retirement and death.
• Beyond θ, q cannot be lowered by raising the wage rate.
𝑤 𝑤
• As goes to 0, q′( ) goes to (negative) infinity.
𝑤𝑅 𝑤𝑅

• It is convenient to assume that q is twice differentiable

𝑤 𝑤
and, for all < θ ; q′′( )>0.
𝑤𝑅 𝑤𝑅

• These assumptions ensure the existence of a unique

solutions for (w, 𝐿𝑀 ) for the firm for every 𝑤𝑅 .
Equilibrium in the Rural and urban labour markets

• Profit is maximised at point N, and firm

෩
should pay wage of 𝑊and employ O𝐿෪
𝑀

workers.
• To find the equilibrium rural employment,
we have use rectangular hyperbola through
the urban employment point.
• The point to note is that gap between the
urban and rural wages arises endogenously.
• That is it is quite possible that w > 𝑤𝑅
without any assumption of wage rigidity
due to political/institutional or trade union
factors.
• The equilibrium wage in manufacturing w is greater than rural wage w.
• It is true that there are people willing to work at wage lower than w, and
employing them would reduce the total wage bill.
• But that gain will be lower than the higher turn over cost due to lower
wage, so employer would not employ people at wage lower than w.
 Labour turnover model: Extensions and Critique
• Low wage with enforceable contract
One possible criticism can be that firms can employ labourers under the enforceable contract that they will not
quit. If there are people willing to work at lower wages, then the firm could offer lower wages and demand a
guarantee of “no quitting”. At equilibrium, firms would make the terms unattractive up to the point where labour
market just clears. However, employment with enforceable no quitting contract are very rare in reality. So we
won’t take this criticism seriously.
• The second issue is that we have assumed that w > 𝒘𝑹
Now the question is what conditions on the primitives of the model ensures this? Consider the following condition

𝒘 𝟏 𝟏
on the quit function and turn over cost w=𝒘𝑹 → 𝒒′ <−
𝒘𝑹 𝒘𝑹 𝑻

𝒘
𝝏𝑻𝒒
𝒘𝑹
The above condition is equivalent w=𝒘𝑹 → < −𝟏
𝝏𝑾
• The above condition says that if w = 𝒘𝑹 , then the slope of the graph in the left panel of
the figure, is less than -1 (note that slopes are given in negative numbers, and also note the
change in the orientation of that graph).
• Since we have assumed that the set above the graph is convex [q′′(w/ 𝒘𝑹 ) > 0], it is clear
that slope cannot be equal to -1 where w < 𝒘𝑹 . Also note that firm’s optimal wage is
where the slope of the turnover cost function equals -1. This ensures that 𝑤
෥ > 𝒘𝑹 .
• Though it solves the direction of the rural-urban wage difference, it is only a technical
condition. Now the question is, is there any conceptual justification for using the above
condition or some alternative explanation of why urban wage exceeds the rural wage.
• One important point is that the model does not explicitly model workers’ decision making.
It is not explained with rigour what motivates a worker’s decision to quit. In the
background it keeps a picture like the following.
• Workers quit their job in search of better one, and a search is necessary because a typical
economy is characterised by an array of wages about which workers have very incomplete
information. Keeping the above view on why workers quit, assume that the firm is paying a
wage lower than 𝒘𝑹 .
• This implies that working for the firm is the worst option open to a worker and firms may
face very high rate of turnover. The above condition along with other assumptions in the
model, may be approximately capturing this idea.
• Further, the above idea of turnover rate brings a dichotomy between
sector and a firm. In the model quitted workers are immediately
replaced by new workers and this enabled the model to consider
turnover rate and quit rate equivalently.
• But when w < 𝑤𝑅 , turnover is not possible, because all workers would
work in the rural sector.
 Labour turnover and Duopsony

Consider an economy consisting of only two firms, 1 and 2. Also assume that labour turnover faced by
firm i is a function of the ratio 𝑤𝑖 and the average wage namely (w1 + w2)/2. For simplicity, we assume
𝑤𝑖
that annual labour turnover, qi faced by firm i is a function of , where i≠ 𝑗. That is
𝑤𝑗

𝒘𝒊
𝒒𝒊 = 𝒒 , 𝒊 ≠ 𝒋, 𝒊 = 𝟏, 𝟐
𝒘𝒋

As in the previous section, the firm chooses wi so as to minimize per worker cost:

𝒘𝒊
𝒘𝒊 = T𝒒
𝒘𝒋
• In this model involuntary unemployment and wage rigidities might arise in
equilibrium.
• For instance, by offering a higher wage, the firm i can reduce the labour turnover and
thereby the cost associated with it, even though there are workers who are willing to
work at lower wages.
• Thus wages will be pushes up to the point where additions to the wage component of
the cost are no longer offset by the diminishing turnover cost.
• In this model also “quit rate” is equated to “labour turnover”, that is each quit is
immediately replaced by new labourer.
• To further illustrate the above point, assume that quit function has the following form

𝒘𝒊
𝒒𝒊 = 𝒊 = 𝟏, 𝟐
𝒘𝒋
• Let T =1. Hence per worker cost takes the following form 𝒘𝒊 + 𝒘𝒋 / 𝒘𝒊
• Firm i chooses 𝒘𝒊 to minimize this and thus we have following first order condition
1 − 𝒘𝒋 /𝒘𝟐𝒊 = 0
• Note that second order condition satisfied.
• Rearranging the first order condition gives firms’ reaction function.
• The reaction functions of firm 1 and 2 are as follows: 𝒘𝟏 = 𝑤2 and 𝒘𝟐 = 𝑤1
• By solving this we get w1 = w2 = 1, or the equilibrium wage configuration, which is a Nash equilibrium.
• A wage configuration (𝒘∗𝟏 , 𝒘∗𝟐 ) is a Nash equilibrium if and only if the following is true: if firm 2 sets its wage
at 𝒘∗𝟐 , then 𝒘∗𝟏 is optimal for firm 1, and if firm 1 sets its wage at 𝒘∗𝟏 , then 𝒘∗𝟐 is optimal for firm 2. That is, it is
a wage configuration from which no agent can benefit by deviating.
• Given that [1,1] is the equilibrium wage vector and substituting this into the labour cost equation
shows that per worker cost faced by both firms at equilibrium is 2.
• The point to note is that at equilibrium both firms are worse off than they would be if they both
agreed to pay a smaller wage.
1
• For instance if both agreed to pay a wage of , then for each firm the cost of employing one
16

1
labour would be 1 , which is better than the situation in equilibrium.
16

1 1
• However, the vector , is not a Nash equilibrium and firms would cut down their cost by
16 16

raising the wage along the reaction curve and ultimately converge to the equilibrium outcome of
[1,1], even if there is unemployment.
• For instance, assume the aggregate labour supply is inelastic and equal to 3 and marginal
2
product of labour of firm i is given by 𝑀𝑃𝑖 = , 𝑖 = 1,2, where 𝐿𝑖 is the amount of labour
𝐿𝑖

employed by the firm i.

• The equilibrium employment is equal to 1 and the aggregate employment is 2 ( =1+1),

resulting in an unemployment of magnitude of 1.