The technological determinants of long‐run inequality

• Published date: 01 April 2018
• DOI: http://doi.org/10.1111/jpet.12248
• Date: 01 April 2018
• Author: Andrea Canidio

Received: 24 August 2016 Accepted: 25 October 2016

DOI: 10.1111/jpet.12248

ARTICLE

The technological determinants of long-run

inequality

Andrea Canidio

INSEAD,Economics and Political Science Area

AndreaCanidio, INSEAD, Economics and

PoliticalScience Area, Boulevard de Con-

stance,77300 Fontainebleau, France

(andrea.canidio@insead.edu).

Iam thankful to Dilip Mookherjee for his precious

helpand support. I am also indebted to Santanu

Chatterjee,Andrew Newman, the participants

tothe microeconomic reading group at ECARES-

ULB,and three anonymous referees for helpful

discussionsand constructive comments.

I explore the effect of skill-biased technological change and unbi-

ased technological progress on long-run inequality using a theoret-

ical model in which the supply of skilled and unskilled workers is

endogenous. The main assumption of the model is that young agents

can finance their education and become skilled workers by borrow-

ing against their future income on an imperfect credit market. I show

that whenever the rate of unbiased technological progress is suffi-

ciently high there is no steady-state inequality, independent of the

degree of skill bias. If instead the rate of unbiased technological

progress is low, then the long-run skill premium increases with the

technological skill bias. Therefore, similarly to the short run, in the

long run higher technological skill bias may cause higher inequality.

However, contrary to the short run, in the long run unbiased tech-

nological progress is more important than technological skill bias in

determining inequality. I also discuss how the efficiency of the edu-

cational technology and the degree of financial development affect

long-run inequality.

1INTRODUCTION

The evolution of the wage structure in the United States between the end of the 1970s and the beginning of the 1990s

suggests that technology can increase short-run inequality. Followingthe introduction of the personal computer and

the unfolding of the information technology era, the difference between the average wage of workerswith a college

degree and of workers with a high school degree increased significantly. This waveof innovations was skill-biased:it

increased the productivity of skilled workers (workers with a college degree), leaving unchanged the productivity of

unskilled workers.1

However,the long-run impact of technology on inequality is not well understood. The reason is that, in the long run

the supply of skilled workers may react to variations in wages. Forexample, parents may be willing to spend more on

the education of their children when the return on education is higher. In addition, the short-run cost of education is

fixed, but the long-run college tuitions are likely to be correlated with the skilled wage because college professors are

1Forempirical evidence, see, among many others, Autor,Katz, and Kearney (2008), Heckman, Lochner, and Taber(1998), and Juhn, Murphy,and Pierce (1993).

156 c

2017 Wiley Periodicals,Inc. wileyonlinelibrary.com/journal/jpet Journal of Public Economic Theory.2018;20:156–176.

CANIDIO 157

skilled workers.Finally, a sufficiently well developed financial system may allow students to borrow against their future

income. When this income is higher,more people should be able to borrow and access education.

The goal of this paper is to explore theoretically the effect of technology on long-run inequality bybuilding an over-

lapping generations model in which the demand and, most importantly,the supply of different types of skills is endoge-

nous. In the model, parents care about the future earnings of their offspring, and leavebequests that are used by young

adults to access education either directly or by first borrowing on the credit market. In addition, the cost of education

is endogenous and is proportional to the wage of skilled workers. Regarding the demand for skills, the model is fairly

standard, in the sense that a competitive production sector pays workerstheir marginal product.

The central assumption of the model is that the credit market is imperfect. Youngagents can borrow against their

future income. However,because of credit market frictions, only agents with a sufficient level of inherited wealth can

access the credit market. It follows that changes in the wage structure affect the number of people who can access the

credit market both directly via the credit market,and indirectly via the equilibrium level of bequests left by parents.

In the model, the degree of altruism of each parent is stochastic, so that there is alwaysa positive probability that an

unskilled parent has a skilled child (and vice versa).Nonetheless, if skilled and unskilled parents earn different amounts

(i.e., there is inequality), the cost of leaving sufficient bequests so that a child can access education will be higher for an

unskilled parent than for a skilled parent. The reason is that, in terms of forgone utility from consumption, the cost of

educating a child is higher for poor parents than for rich parents. This observation implies that inequality is possible in

steady state, because the higher the difference in earnings between skilled and unskilled workers, the lower is social

mobility,which in turn implies that skilled workers may be scarce and earn a wage premium over unskilled workers.

The above reasoning fails—and inequality is not possible in steady state—whenever all young individuals can bor-

row on the credit market and finance their education, independent of the inheritance received. Hence, steady-state

inequality can exist only if credit rationing exists, so that only agents born with wealth abovea certain threshold can

become skilled workers. The main result of the paper is to show that the existence of credit rationingdepends on the

level of development of the financial sector,on the efficiency of the educational sector, and on the unbiased growth

rate of the economy. Crucially, skill-biased technological change plays no role in the existence of credit rationing. In

this sense, whereas the unbiased growth rate of the economy can be considered a first-order determinant of long-run

inequality,skill-biased technological change is, at best, a second-order determinant.

Intuitively,the effect of an increase in skill bias on the functioning of the credit market has two components. Higher

skill bias increases both the cost of education and the future skilled wage, which can be used as collateral in the credit

market. When the efficiency of the schooling technology (which determines the cost of education), the development

of the financial sector (which determines the presence of credit market frictions), and the growth rateof the economy

(which determines the future skilled wage) are high enough, the positive effect always dominates. As a consequence,

for any levelof skill bias a young agent who is born with zero wealth can access the credit market and become a skilled

worker,because the future wage itself provides enough collateral to access the credit market. The converseis also true,

because if a young agent who is born with no wealth can access the credit market for some levelof skill bias, then this

agent can access the credit market for anylevel of skill bias.

If credit market rationing is present because, for example,the growth rate of the economy is low, then higher tech-

nological skill bias increases long-run inequality.A higher skill bias increases the likelihood that the steady state of the

economy is unequal, and increases the steady-state skill premium if the steady state is unequal. Intuitively,the cost of

leaving bequests large enough so that a child can access education is always greater for an unskilled parent than for a

skilled parent, the more so the higher is the skill bias. Hence, higher skill bias decreases intergenerationalmobility. This

effect, coupled with an increase in the skill premium, generates higher long-run inequality following an increase in the

skill bias.

Thesame framework can be applied to understanding the long-run effects of other technological and policy changes.

For example, the efficiency of the educational technology is a first-order determinant of long-run inequality, in the

sense that when it is sufficiently high there is no inequality in steady state. Furthermore,in case steady-state inequality

exists,an increase in the efficiency of the educational technology decreases the level of steady-state inequality.There-

fore, the model suggests that innovationin the educational sector (i.e., the introduction of online learning in the form of