Economic As 6. Economic Aspects of Automation pextras.springer.com/2009/978-3-540-78830-0/... · Economic As 6. Economic Aspects of Automation p Piercarlo Ravazzi, Agostino Villa

93

Economic Asp6. Economic Aspects of Automation

Piercarlo Ravazzi, Agostino Villa

The increasing diffusion of automation in all sec-tors of the industrial world gives rise to a deepmodification of labor organization and requiresa new approach to evaluate industrial systemsefficiency, effectiveness, and economic conve-nience. Until now, the evaluation tools andmethods at disposal of industrial managers arerare and even complex. Easy-to-use criteria, pos-sibly based on robust but simple models andconcepts, appear to be necessary. This chaptergives an overview of concepts, based on the eco-nomic theory but revised in the light of industrialpractice, which can be applied for evaluating theimpact and effects of automation diffusion inenterprises.

6.1 Basic Conceptsin Evaluating Automation Effects ........... 96

6.2 The Evaluation Model ........................... 976.2.1 Introductory Elements

of Production Economy ................. 976.2.2 Measure of Production Factors ....... 976.2.3 The Production Function

Suggested by Economics Theory ..... 98

6.3 Effects of Automation in the Enterprise .. 986.3.1 Effects of Automation

on the Production Function ........... 986.3.2 Effects of Automation

on Incentivizationand Control of Workers ................. 100

6.3.3 Effects of Automationon Costs Flexibility ....................... 101

6.4 Mid-Term Effects of Automation ............ 1026.4.1 Macroeconomics Effects

of Automation:Nominal Prices and Wages ............ 102

6.4.2 Macroeconomics Effectsof Automation in the Mid-Term:Actual Wagesand Natural Unemployment .......... 105

6.4.3 Macroeconomic Effectsof Automation in the Mid Term:Natural Unemploymentand Technological Unemployment.. 107

6.5 Final Comments.................................... 111

6.6 Capital/Labor and Capital/Product Ratiosin the Most Important Italian IndustrialSectors ................................................ 113

References .................................................. 115

Process automation spread through the industrial worldin both production and services during the 20th century,and more intensively in recent decades. The condi-tions that assured its wide diffusion were first thedevelopment of electronics, then informatics, and todayinformation and communication technologies (ICT), asdemonstrated in Fig. 6.1a–c. Since the late 1970s, pe-riods of large investment in automation, followed byperiods of reflection with critical revision of previ-ous implementations and their impact on revenue, havetaken place. This periodic attraction and subsequent

revision of automation applications is still occurring,mainly in small to mid-sized enterprises (SME) as wellas in several large firms.

Paradigmatic could be the case of Fiat, whichreached the highest level of automation in their assem-bly lines late in the 1980s, whilst during the subsequentdecade it suffered a deep crisis in which investmentsin automation seemed to be unprofitable. However, thenext period – the present one – is characterized by sig-nificant growth for which the high level of automationalready at its disposal has been a driver.

PartA

6

94 Part A Development and Impacts of Automation

0 20 40 60 80 100

a)

Business with 10 or more employees (%)

Iceland (a)Finland

Greece (a)Ireland

PortugalDenmark

Netherlands (a)NorwayPoland

Slovak RepublicSwedenAustria

Switzerland (b,1)Australia (a)

New Zealand (a)HungaryGermany

EU15Korea (b,1)

United KingdomSpain

BelgiumLuxembourg

ItalyCzech Republic

43

0 20

95

40 60 80 100

b)

Business with 10 or more employees (%)

Iceland (a)Finland

Slovak RepublicJapan (a)

Switzerland (b)Denmark

AustriaKorea (a)

Netherlands (a)BelgiumNorway

GermanyCzech RepublicOECD average

Canada (a)Sweden

New Zealand (a) France (a)

LuxembourgSpainItaly

Ireland (1)Greece (a)

Australia (a)United Kingdom (2)

PolandMexico (d)

PortugalHungary

Automation implementation and perception of itsconvenience in the electronics sector is different from inthe automotive sector. The electronics sector, however,

Fig. 6.1 (a) Use of information and communication tech-nologies by businesses for returning filled forms to publicauthorities. The use of the Internet emphasizes the impor-tant role of transaction automation and the implementationof automatic control systems and services (source: OECD,ICT database, and Eurostat, community survey on ICTusage in households and by individuals, January 2008).(b) Business use of the Internet, 2007, as a percentageof businesses with ten or more employees. (c) � Internetselling and purchasing by industry (2007). Percentage ofbusinesses with ten or more employees in each industrygroup (source: OECD, ICT database, and Eurostat, com-munity survey on ICT usage in enterprises, September2008)

is rather specific since it was born together with – andas the instrument of – industrial automation. All otherindustrial sectors present a typical attitude of strongbut cautious interest in automation. The principal mo-tivation for this caution is the difficulty that managersface when evaluating the economic impact of automa-tion on their own industrial organization. This difficultyresults from the lack of simple methods to estimate eco-nomic impact and obtain an easily usable measure ofautomation revenue.

The aim of this chapter is to present an evalua-tion approach based on a compact and simple economicmodel to be used as a tool dedicated to SME managers,to analyze main effects of automation on production,labor, and costs.

The chapter is organized as follows. First, somebasic concepts on which the evaluation of the automa-tion effects are based are presented in Sect. 6.2. Then,a simple economic model, specifically developed foreasy interpretation of the impact of automation on theenterprise, is discussed and its use for the analysis ofsome industrial situations is illustrated in Sect. 6.3. Themost important effects of automation within the enter-prise are considered in Sect. 6.4, in terms of its impacton production, incentivization and control of workers,and costs flexibility. In the final part of the chapter,mid-term effects of automation in the socioeconomiccontext are also analyzed. Considerations of such ef-fects in some sectors of the Italian industrial systemare discussed in Sect. 6.6, which can be considered asa typical example of the impact of automation on a de-veloped industrial system, easily generalizable to othercountries.

PartA

6

Economic Aspects of Automation 95

0 20 40 60 8080Selling Purchasing 60 40 20

c)

Aus

tralia

(3) Construction

ManufacturingReal estate, renting & business activities

Transport & storageWholesale

RetailAll industries

ConstructionManufacturing

Real estate, renting & business activitiesTransport, storage & communication

WholesaleRetail

All industries

Can

ada



Whole saleAll industries



Wholesale tradeAll industries



Wholesale & retailAll industries

Finl

and



Wholesale & retailAll industriesG

erm

any




Gre

ece




Hun

gary




Icel

and

(200

6)




Irel

and




Japa

nIta

ly


Real estate, renting & business activitiesTransport & storageWholesale & retail

All industries

Aus

tria

Bel

gium

Cze

chR

epub

lic ConstructionManufacturing


Wholesale tradeAll industries

Den

-m

ark Construction


PartA

6


6.1 Basic Concepts in Evaluating Automation Effects

The desire of any SME manager is to be able to eval-uate how to balance the cost of implementing someautomated devices (either machining units or handlingand moving mechanisms, or automated devices to im-prove production organization) and the related increaseof revenue.

To propose a method for such an economic evalua-tion, it is first necessary to declare a simple catalogueof potential automation typologies, then to supply evi-dence of links between these typologies and the mainvariables of a SME which could be affected by processand labor modifications due to the applied automa-tion. All variables to be analyzed and evaluated mustbe the usual ones presented in a standard balancesheet.

Analysis of a large number of SME clusters inten European countries, developed during the collab-orative demand and supply networks (CODESNET)project [6.1] funded by the European Commission,shows that the most important typologies of automa-tion implementations in relevant industrial sectors canbe classified as follows:

1. Robotizing, i. e., automation of manufacturingoperations

2. Flexibilitization, i. e., flexibility through automation,by automating setup and supply

3. Monitorizing, i. e., monitoring automation throughautomating measures and operations control.

These three types of industrial automation can be re-lated to effects on the process itself as well as onpersonnel. Robotizing allows the application of greateroperation speed and calls for a reduced amount of di-rect work hours. Flexibilitization is crucial in masscustomization, to reduce the lead time in the face

Table 6.1 Links between automation and process/personnel in the firm

Automation typology induces . . . . . . effects on the process . . . . . . and effects on personnel

(a) Robotizing Operation speed Work reduction

(b) Flexibilitization Response time to demand Higher skills

(c) Monitorizing Process accuracy and product quality

Then automation calls for . . . . . . and search for . . .

Investments New labor positions

Investments and new labour positions should give rise to an expected target of production, conditioned on investments

in high technologies and highly skill workforce utilization.

of customer demands, by increasing the product mix,and by facilitating producer–client interaction. Moni-torizing can indeed assure product quality for a widerange of final items through diffused control of workoperations. Both automated flexibility and automatedmonitoring, however, require higher skills of personnel(Table 6.1).

However, a representation of the links betweenautomation and either process attributes or personnelworking time and skill, as outlined in Table 6.1, does notcorrespond to a method for evaluating the automation-induced profit in a SME, or in a larger enterprise. It onlyshows effects, whereas their impact on the SME balancesheet is what the manager wants to know.

To obtain this evaluation it is necessary:

1. To have clear that an investment in automation isgenerally relevant for any enterprise, and often crit-ical for a SME, and typically can have an impact onmid/long-term revenue.

2. To realize that the success of an investment in termsof automation depends both on the amount of in-vestment and on the reorganization of the workforcein the enterprise, which is a microeconomic effect(meaning to be estimated within the enterprise).

3. To understand that the impact of a significant invest-ment in automation, made in an industrial sector,will surely have long-term and wide-ranging effectson employment at the macroeconomic level (i. e., atthe level of the socioeconomic system or country).

All these effects must be interpreted by a singleevaluation model which should be used:

• For a microeconomic evaluation, made by theenterprise manager, to understand how the two

PartA

6.1

Economic Aspects of Automation 6.2 The Evaluation Model 97

above-mentioned principal factors, namely invest-ment and workforce utilization, could affect theexpected target of production, in case of a givenautomation implementation• For a macroeconomic evaluation, to be done at thelevel of the industrial sector, to understand how rel-evant modification of personnel utilization, caused

by the spread of automation, could be reflected inthe socioeconomic system.

These are the two viewpoints according to which theautomation economic impact will be analyzed uponthe introduction of the above-mentioned interpretationmodel in Sect. 6.2.

6.2 The Evaluation Model

6.2.1 Introductory Elementsof Production Economy

Preliminary, some definitions and notations from eco-nomics theory are appropriate (see the basic refer-ences [6.2–6]):

• A production technique is a combination of factorsacquired by the market and applied in a prod-uct/service unit.• Production factors will be simply limited to the cap-ital K (i. e., industrial installation, manufacturingunits, etc.), to labor L , and to intermediate goods X(i. e., goods and services acquired externally to con-tribute to production).• A production function is given by the relationQ = Q(K, L, X), which describes the output Q,production of goods/services, depending on the ap-plied inputs.• Technological progress, of which automation is themost relevant expression, must be incorporated intothe capital K in terms of process and labor innova-tions through investments.• Technical efficiency implies that a rational manager,when deciding on new investments, should makea choice among the available innovations that allowhim to obtain the same increase of production with-out waste of inputs (e.g., if an innovation calls forK0 units of capital and L0 units of labor, and an-other one requires L1 > L0 units of labor, for thesame capital and production, the former is to bepreferred).• Economic efficiency imposes that, if the combina-tion of factors were different (e.g., for the sameproduction level, the latter innovation has to useK1 < K0 capital units), then the rational manager’schoice depends on the cost to be paid to implementthe innovation, thus accounting also for productioncosts, not only quantity.

Besides these statements it has to be remarked that,according to economic theory, production techniquescan be classified as either fixed-coefficients technolo-gies or flexible-coefficients technologies. The formerare characterized by nonreplaceable and strictly com-plementary factors, assuming that a given quantity ofproduction can only be obtained by combining produc-tion factors at fixed rates, with the minimum quantitiesrequired by technical efficiency. The latter are charac-terized by the possibility of imperfect replacement offactors, assuming that the same production could beobtained through a variable, nonlinear combination offactors.

6.2.2 Measure of Production Factors

Concerning the measure of production factors, the fol-lowing will be applied.

With regard to labor, the working time h (hours)done by personnel in the production system, is as-sumed to be a homogeneous factor, meaning thatdifferent skills can be taken into account through suit-able weights.

In case of N persons in a shift and T shifts in unittime (e.g., day, week, month, etc.), the labor quantity Lis given by

L = hNT . (6.1)

In the following, the capital K refers to machines andinstallations in the enterprise. However, it could alsobe easily measured in the case of fixed-coefficient tech-nologies: the capital stock, indeed, could be measured interms of standard-speed machine equivalent hours. Thecapital K can also be further characterized by notingthat, over a short period, it must be considered a fixedfactor with respect to production quantity: excess capac-ity cannot be eliminated without suffering heavy losses.

Labor and intermediate goods should rather be vari-able factors with respect to produced quantities: excess

PartA

6.2


stock of intermediate goods can be absorbed by re-ducing the next purchase, and excess workforce couldbe reduced gradually through turnover, or suddenlythrough dismissals or utilization of social measures infavor of unemployees.

In the long term, all production factors should beconsidered variables.

6.2.3 The Production FunctionSuggested by Economics Theory

The above-introduced concepts and measures allow tostate the flexible-coefficients production function, by as-suming (for the sake of simplicity and realism) thatonly the rate of intermediate goods over production isconstant (X/Q = b)

Q = Q(K, L, X) = f (K, L) = X/b . (6.2)

The production function is specified by the followingproperties:

• Positive marginal productivity (positive variation ofproduction depending on the variation of a singlefactor, with the others being fixed, i. e., ∂Q/∂K > 0,∂Q/∂L > 0)• Decreasing returns, so that marginal productivity isa decreasing function with respect to any productionfactor, i. e., ∂Q2/∂K2 < 0, ∂Q2/∂L2 < 0.

In economics theory, from Clark and Marshall untilnow, the basis of production function analysis was thehypothesis of imperfect replacement of factors, assign-ing to each the law of decreasing returns.

The first-generation approach is due to Cobb andDouglas [6.7], who proposed a homogeneous pro-duction function whose factors can be additive inlogarithmic form: Q = ALa Kb, where the constant Asummarizes all other factors except labor and capital.

This formulation has been used in empirical investiga-tions, but with severe limitations.

The hypothesis of elasticity from the Cobb–Douglas function with respect to any production factor[such that (∂Q/Q)/(∂L/L) = (∂Q/Q)/(∂K/K ) = 1],has been removed by Arrow et al. [6.8] and Brownand De Cani [6.9]. Subsequent criticism by McFad-den [6.10] and Uzawa [6.11] gave rise to the moregeneral form of variable elasticity function [6.12],up to the logarithmic expression due to Christensenet al. [6.13, 14], as clearly illustrated in the system-atic survey due to Nadiri [6.15]. The strongest criticismon the flexible coefficient production function has beenprovided by Shaikh [6.16], but seems to have been ig-nored. The final step was that of abandoning directestimation of the production function, and applying in-direct estimation of the cost function [6.2,17–20], up tothe most recent theories of Dievert [6.21] and Jorgen-son [6.22].

A significant modification to the analysis approachis possible based on the availability of large statisticaldatabases of profit and loss accounts for enterprises,compared with the difficulty of obtaining data con-cerning production factor quantities. This approachdoes not adopt any explicit interpretation scheme,thus upsetting the approach of economics theory (de-ductive) and engineering (pragmatic), depending onlyon empirical verification. A correct technological–economic approach should reverse this sequence: withreference to production function analysis, it shouldbe the joint task of the engineer and economist topropose a functional model including the typical param-eters of a given production process. The econometrictask should be applied to verify the proposed modelbased on estimation of the proposed parameters. Inthe following analysis, this latter approach will beadopted.

6.3 Effects of Automation in the Enterprise

6.3.1 Effects of Automationon the Production Function

The approach on which the following considerationsare based was originally developed by Luciano andRavazzi [6.23], assuming the extreme case of produc-tion only using labor, i. e., without employing capital(e.g., by using elemental production means). In thiscase, a typical human characteristic is that a worker canproduce only at a rate that decreases with time during

his shift. So, the marginal work productivity is decreas-ing.

Then, taking account of the work time h of a workerin one shift, the decreasing efficiency of workers withtime suggests the introduction of another measure,namely the efficiency unit E, given by

E = hα , (6.3)

where 0 < α < 1 is the efficiency elasticity with respectto the hours worked by the worker.

PartA

6.3

Economic Aspects of Automation 6.3 Effects of Automation in the Enterprise 99

Condition (6.3) includes the assumption of decreas-ing production rate versus time, because the derivativeof E with respect to h is positive but the second deriva-tive is negative.

Note that the efficiency elasticity can be viewed asa measure of the worker’s strength.

By denoting λE as the production rate of a workunit, the production function (6.2) can be rewritten as

Q = λE ENT . (6.4)

Then, substitution of (6.1) and (6.3) into (6.4), givesrise to a representation of the average production rate,which shows the decreasing value with worked hours

λL = Q/L = λEhα−1 , (6.5)

with dλL/dh = (α−1)λEhα−2 < 0.Let us now introduce the capital K as the auxiliary

instrument of work (a computer for intellectual work,an electric drill for manual work, etc.), but without anyprocess automation.

Three questions arise:

1. How can capital be measured?2. How can the effects produced by the association of

capital and work be evaluated?3. How can capital be included in the production func-

tion?

With regard to the first question, the usual industrialapproach is to refer to the utilization time of the pro-duction instruments during the working shift. Then, letthe capital K be expressed in terms of hours of poten-tial utilization (generally corresponding to the workingshift).

Utilization of more sophisticated tools (e.g., throughthe application of more automation) induces an increasein the production rate per hour. Denoting by γ > 1 a co-efficient to be applied to the production rate λL, in orderto measure its increase due to the effect of capital uti-lization, the average production rate per hour (6.5) canbe rewritten as

λL = γλEhα−1 . (6.6)

The above-mentioned effect of capital is the only oneconsidered in economics theory (as suggested by theCobb–Douglas function). However, another significanteffect must also be accounted for: the capital’s impacton the workers’ strength in terms of labor (as mentionedin the second question above).

Automated systems can not only increase pro-duction rate, but can also strengthen labor efficiencyelasticity, since they reduce physical and intellectual

fatigue. To take account of this second effect, condi-tion (6.6) can be reformulated by including a positiveparameter δ > 0 that measures the increase of labor ef-ficiency and whose value is bounded by the condition0 < (α+ δ) < 1 so as to maintain the hypothesis of de-creasing production rate with time

λL = γλEhα+δ−1 . (6.7)

According to this model, a labor-intensive techniqueis defined as one in which capital and labor cooper-ate together, but in which the latter still dominates theformer, meaning that a reduction in the labor marginalproduction rate still characterizes the production pro-cess (α+ δ < 1), even if it is reduced by the capitalcontribution (δ > 0).

The answer to the third question, namely how toinclude capital in the production function, strictly de-pends on the characteristics of the relevant machinery.Whilst the workers’ nature can be modeled based on theassumption of decreasing production rates with time,production machinery does not operate in this way (onecould only make reference to wear, although mainte-nance, which can prevent modification of the productionrate, is reflected in capital cost).

On the contrary, it is the human operator who im-poses his biological rhythm (e.g., the case of the speedof a belt conveyor that decreases in time during theworking shift). This means that capital is linked to pro-duction through fixed coefficients: then the marginalproduction rate is not decreasing with the capital.

Indeed, a decreasing utilization rate of capital hasto be accounted for as a consequence of the decreas-ing rate of labor. So, the hours of potential utilizationof capital K have to be converted into productivehours through a coefficient of capital utilization θ andtransformed into production through a constant capital-to-production rate parameter v

Q = θK/v = θλK K , (6.8)

where λK = 1/v is a measure of the capital constantproductivity, while 0 < θ < 1 denotes the ratio betweenthe effective utilization time of the process and the timeduring which it is available (i. e., the working shift).

Dividing (6.8) by L and substituting into (6.7), itfollows that

θ = vγλEhα+δ−1 , (6.9)

thus showing how the utilization rate of capital could fitthe decreasing labor yield so as to link the mechanicalrhythm of capital to the biological rhythm of labor.

Condition (6.9) leads to the first conclusion: inlabor-intensive systems (in which labor prevails over

PartA

6.3


capital), decreasing yields occur, but depending only onthe physical characteristics of workers and not on theconstant production rates of production machinery.

We should also remark on another significant con-sideration: that new technologies also have the functionof relieving labor fatigue by reducing undesirable ef-fects due to marginal productivity decrease.

This second conclusion gives a clear suggestion ofthe effects of automation concerning reduction of phys-ical and intellectual fatigue. Indeed, automation impliesdominance of capital over labor, thus constraining laborto a mechanical rhythm and removing the conditioningeffects of biological rhythms.

This situation occurs when α+ δ = 1, thus modify-ing condition (6.7) to

λL = Q/L = γλE , (6.10)

which, in condition (6.9), corresponds to θ = 1, i. e., nopause in the labor rhythm.

In this case automation transforms the decreasingyield model into a constant yield model, i. e., the laborproduction rate is constant, as is the capital productionrate, if capital is fully utilized during the work shift.Then, capital-intensive processes are defined as thosethat incorporate high-level automation, i. e., α+ δ → 1.

A number of examples of capital-intensive pro-cesses can be found in several industrial sectors, oftenconcerning simple operations that have to be executeda very large number of times. A typical case, even if notoften considered, are the new intensive picking systemsin large-scale automated warehouses, with increasingdiffusion in large enterprises as well as in industrialdistricts.

Section 6.6 provides an overview in several sec-tors of the two ratios (capital/labor and produc-tion/labor) that, according to the considerations above,can provide a measure of the effect of automationon production rate. Data are referred to the Italianeconomic/industrial system, but similar considerationscould be drawn for other industrial systems in devel-oped countries. Based on the authors’ experience duringthe CODESNET project development, several Euro-pean countries present aspects similar to those outlinedin Sect. 6.6.

6.3.2 Effects of Automationon Incentivizationand Control of Workers

Economic theory recognizes three main motivationsthat suggest that the enterprise can achieve greater wageefficiency than the one fixed by the market [6.24]:

1. The need to minimize costs for hiring and trainingworkers by reducing voluntary resignations [6.25,26]

2. The presence of information asymmetry betweenthe workers and the enterprise (as only workersknow their ability and diligence), so that the en-terprise tries to engage the best elements from themarket through ex ante incentives, and then to forcequalified employees to contribute to the productionprocess (the moral hazard problem) without result-ing in too high supervision costs [6.27–29]

3. The specific features of production technologiesthat may force managers to allow greater autonomyto some worker teams, while paying an incentivein order to promote better participation in team-work [6.30–32].

The first motivation will be discussed in Sect. 6.4,concerning the flexibility of labor costs, while the lastone does not seem to be relevant. The second motivationappears to be crucial for labor-intensive systems, sincesystem productivity cannot only be dependent on tech-nologies and workers’ physical characteristics, but alsodepends greatly on workers propensity to contribute.

So, system productivity is a function of wages, andmaximum profit can no longer be obtained by apply-ing the economic rule of marginal productivity equal towages fixed by market. It is the obligation of the en-terprise to determine wages so as to assure maximumprofit.

Let the production rate of a work unit λE increaseat a given rate in a first time interval in which the wagewE per work unit plays a strongly incentive role, whilstit could increase subsequently at a lower rate owing tothe reduction of the wages marginal utility, as modeledin the following expression, according to the economichypothesis of the effort function

λE = λE(wE), where λE(0) = 0; dλE/dwE > 0 ;and d2λE/dw2

E ≥ 0 if wE ≥ w ;d2λE/dw2

E ≤ 0 if wE ≤ w ; (6.11)

where w is the critical wages which forces a change ofyield from increasing to decreasing rate.

In labor-intensive systems, the average productionrate given by (6.7) can be reformulated as

λL = γλE E/h . (6.12)

Now, let M = Va −wL be the contribution margin, i. e.,the difference between the production added value Vaand the labor cost: then the unitary contribution margin

PartA

6.3

Economic Aspects of Automation 6.3 Effects of Automation in the Enterprise 101

per labor unit m is defined by the rate of M over thelabor L

m = M/L = pλL −w , (6.13)

where p = (p− pX)β is the difference between the saleprice p and the cost pX of a product’s parts and mater-ials, transformed into the final product according to theutilization coefficient β = X/Q.

It follows that pλL is a measure of the added value,and that the wages wE per work unit must be trans-formed into real wages through the rate of work units Eover the work hours h of an employee during a workingshift, according to

w = wE E/h . (6.14)

The goal of the enterprise is to maximize m, in order togain maximum profit, i. e.,

max(m) = [pγλE(wE)−wE

]E/h . (6.15)

The first-order optimal condition gives

∂m/∂h = [pγλE(wE)−wE

]∂(E/h)/∂h = 0

⇒ γλE(wE) = wE/ p , (6.16)

∂m/∂wE = (E/h)[

pγ (∂λE/∂wE)−1] = 0

⇒ pγ (∂λE/∂wE) = 1 . (6.17)

By substituting (6.17) into (6.16), the maximum-profitcondition shows that the elasticity of productivity withrespect to wages ελ will assume a value of unity

ελ = (∂λE/∂wE)(wE/λE) = 1 . (6.18)

So, the enterprise could maximize its profit by forc-ing the percentage variation of the efficiency wages tobe equal to the percentage variation of the productiv-ity ∂λE/λE = ∂wE/wE. If so, it could obtain the optimalvalues of wages, productivity, and working time.

As a consequence, the duration of the working shiftis an endogenous variable, which shows why, in labor-intensive systems, the working hours for a worker candiffer from the contractual values.

On the contrary, in capital-intensive systems withwide automation, it has been noted before that E/h = 1and λE = λE, because the mechanical rhythm prevailsover the biological rhythm of work. In this case, effi-ciency wages do not exist, and the solution of maximumprofit simply requires that wages be fixed at the min-imum contractual level

max(m) = pλL−w = pγ λE −w ⇒ min(w) . (6.19)

As a conclusion, in labor-intensive systems, if λE couldbe either observed or derived from λL, incentive wagescould be used to maximize profit by asking workers foroptimal efforts for the enterprise. In capital-intensivesystems, where automation has canceled out deceasingyield and mechanical rhythm prevails in the productionprocess, worker incentives can no longer be justified.The only possibility is to reduce absenteeism, so thata share of salary should be reduced in case of negli-gence, not during the working process (which is fullycontrolled by automation), but outside.

In labor-intensive systems, as in personal serviceproduction, it could be difficult to measure workers’productivity: if so, process control by a supervisor be-comes necessary.

In capital-intensive systems, automation eliminatesthis problem because the process is equipped with de-vices that are able to detect any anomaly in processoperations, thus preventing inefficiency induced by neg-ligent workers. In practice automation, by forcing fixedcoefficients and full utilization of capital (process), per-forms itself the role of a working conditions supervisor.

6.3.3 Effects of Automationon Costs Flexibility

The transformation of a labor-intensive process intoa capital-intensive one implies the modification of thecost structure of the enterprise by increasing the capitalcost (that must be paid in the short term) while reducinglabor costs.

Let the total cost CT be defined by the costs of thethree factors already considered, namely, intermediategoods, labor, and capital, respectively,

CT = pX X +wL + cK K , (6.20)

where cK denotes the unitary cost of capital.Referring total cost to the production Q, the cost

per production unit c can be stated by substituting theconditions (6.2) and (6.10) into (6.20), and assumingconstant capital value in the short term

c = CT/Q = pXβ +w/λL + cK K/Q . (6.21)

In labor-intensive systems, condition (6.21) can alsobe rewritten by using the efficiency wages w∗

E allo-cated in order to obtain optimal productivity λ∗

L =γλE(w∗

E)h∗α+δ−1, as shown in Sect. 6.3.1,

c = pXβ +w∗E/λ∗

L + cK K/Q . (6.22)

On the contrary, in capital-intensive systems, the pres-ence of large amounts of automation induces thefollowing effects:

PartA

6.3


1. Labor productivity λA surely greater than that whichcould be obtained in labor-intensive systems (λA >

λ∗L)

2. A salary wA that does not require incentives toobtain optimum efforts from workers, but which im-plies an additional cost with respect to the minimumsalary fixed by the market (wA

<>

w∗E), in order to

select and train personnel3. A positive correlation between labor productivity

and production quantity, owing to the presence ofqualified personnel who the enterprise do not like tosubstitute, even in the presence of temporary reduc-tions of demand from the final product market

λA = λA(Q), ∂λA/∂Q > 0, ∂2λA/∂Q2 = 0

(6.23)

4. A significantly greater cost of capital, due to thehigher cost of automated machinery, than that ofa labor-intensive process (cKA > cK), even for thesame useful life and same rate of interest of the loan.

According to these statements, the unitary cost incapital-intensive systems can be stated as

cA = pXβ +wA/λA(Q)+ cKA K/Q . (6.24)

Denoting by profit per product unit π the differencebetween sale price and cost

π = p− c , (6.25)

the relative advantage of automation D, can beevaluated by the following condition, obtained by sub-stituting (6.24) and then (6.22) into (6.25)

D = πA −π

= w∗E/λ∗

L −wA/λA(Q)− (cKA − cK)K/Q .

(6.26)

Except in extreme situations of large underutilizationof production capacity (Q should be greater than thecritical value QC), the inequality w∗

E/λ∗L > wA/λA(Q)

denotes the necessary condition that assures that au-tomated production techniques can be economicallyefficient. In this case, the greater cost of capital can becounterbalanced by greater benefits in terms of laborcost per product unit.

In graphical terms, condition (6.26) could be il-lustrated as a function D(Q) increasing with theproduction quantity Q, with positive values for produc-tion greater than the critical value QC. This means thatlarge amounts of automation can be adopted for high-quantity (mass) production, because only in this casecan the enterprise realize an increase in marginal pro-ductivity sufficient to recover the initial cost of capital.This result, however, shows that automation could bemore risky than labor-intensive methods, since produc-tion variations induced by demand fluctuations couldreverse the benefits. This risk, today, is partly reducedby mass-customized production, where automation andprocess programming can assure process flexibility ableto track market evolutions [6.33].

6.4 Mid-Term Effects of Automation

6.4.1 Macroeconomics Effectsof Automation:Nominal Prices and Wages

The analysis has been centered so far on microe-conomics effects of automation on the firm’s costs,assuming product prices are given, since they would beset in the market depending on demand–offer balance ina system with perfect competition.

Real markets however lack a Walrasian auctioneerand are affected by the incapacity of firms to knowin advance (rational expectations) the market demandcurve under perfect competition, and their own demandcurve under imperfect competition. In the latter case,enterprises cannot maximize their profit on the basisof demand elasticity, as proposed by the economic the-

ory of imperfect competition [6.34–36]. Therefore, theycannot define their price and the related markup on theircosts.

Below we suggest an alternative approach, whichcould be described as technological–managerial.

The balance price is not known a priori and pricesetting necessarily concerns enterprises: they have tosubmit to the market a price that they consider to beprofitable, but not excessive because of the fear thatcompetitors could block selling of all the scheduled pro-duction. Firms calculate the sale price p on the basis ofa full unit cost c, including a minimum profit, consid-ered as a normal capital remuneration.

The full cost is calculated corresponding to a sched-uled production quantity Q e that can be allegedly soldon the market, leaving a small share of productive ca-

PartA

6.4

Economic Aspects of Automation 6.4 Mid-Term Effects of Automation 103

pacity Q potentially unused

Q e ≤ Q = θKλK K = θK K/v , (6.27)

where θK = 1 in the presence of automation, whilev = 1/λK is – as stated above – the capital–product con-nection defining technology adopted by firms for fullproductive capacity utilization.

The difference Q − Q e therefore represents the un-used capacity that the firm plans to leave available forproduction demand above the forecast.

To summarize, the sales price is fixed by the en-terprise, resorting to connection (6.21), relating thebreak-even point to Q e

p = c(Q e) = pXβ +w/λL + cKv e , (6.28)

where v e = K/Q e ≥ v (for Q e ≤ Q) is the pro-grammed capital–product relationship.

In order to transfer this relation to the macroeco-nomics level, we have to express it in terms of addedvalue (Q is the gross saleable production), as the grossdomestic product (GDP) results from aggregation of theadded values of firms. Therefore we define the addedvalue as

PY = pQ − pX X = (p− pXβ)Q ,

where P represents the prices general level (the averageprice of goods that form part of the GDP) and Y theaggregate supply (that is, the GDP). From this relation,P is given by

P = (p− pXβ)/θY , (6.29)

where θY = Y/Q measures the degree of verticalintegration of the economic system, which in themedium/short term we can consider to be steady(θY = θY).

By substituting relation (6.28) into (6.29), the priceequation can be rewritten as

P = w/λ+ cKv , (6.30)

where work productivity and the capital/product ratioare expressed in terms of added value (λ = θYλL = Y/Land v = v e/θY = K/Y e).

In order to evaluate the effects of automation at themacroeconomic level, it is necessary to break up thecapital unit cost cK into its components:

• The initial purchase unit price of capital P0K• The sample gross profit ρ∗ sought by firms (as a per-

centage to be applied to the purchase price), whichembodies both amortization rate d and the perfor-mance r∗ requested by financiers to remunerate

debt (subscribed by bondholders) and risk capital(granted by owners).

So the following definition can be stated

cK = ρ∗ P0K = [

l∗ P0K + (1− l∗)PK

]/an |r∗ . (6.31)

where

• 0 < l∗ = D∗/(P0K K ) < 1 is the leverage (debt

amount subscribed in capital stock purchase)• 1− l∗ is the amount paid by owners• PK > P0K is the substitution price of physical capital

at the deadline and• an |r∗ = 1−(1+r∗)−n

r∗ is the discounting back factor.

Relation (6.31) implies that the aim of the firm is tomaintain unchanged the capital share amount initiallybrought by owners, while the debt amount is recoveredto its face (book) value, as generally obligations are re-funded at original monetary value and at a fixed interestrate stated in the contract.

It is also noteworthy that the indebtedness ratio lis fixed at its optimal level l∗, and the earning ratedepends on it, r∗ = r(l∗), because r decreases as thedebt-financed share of capital increases due to advan-tages obtained from the income tax deductibility ofstakes [6.37, 38], and increases as a result of failurecosts [6.39–41] and agency costs [6.42], which in turngrow as l grows. The optimal level l∗ is obtained basedon the balance between costs and marginal advantages.

The relation (6.31) can be rewritten by using a Tay-lor series truncated at the first term

cK = (d + r∗)[l∗ P0

K + (1− l∗)PK], (6.32)

which, in the two extreme cases P0K = PK and PK = P0

K,can be simplified to

cK = (d + r∗)PK , (6.33a)

cK = (d + r∗)P0K . (6.33b)

This solution implies the existence of monetary illu-sion, according to which capital monetary revaluation(following inflation) is completely abandoned, thus im-peding owners from keeping their capital intact.

This irrational decision is widely adopted in prac-tice by firms when inflation is low, as it rests upon theaccounting procedure codified by European laws thatcalculated amortization and productivity on the basis ofbook value. Solution (6.33a) embodies two alternatives:

• Enterprises’ decision to maintain physical capitalundivided [6.43], or to recover the whole capital

PartA

6.4


market value at the deadline, instead of being re-stricted to the capital value of the stakeholders, inorder to guarantee its substitution without reducingexisting production capacity; in this case the indebt-edness with repayment to nominal value (at fixedrate) involves an extra profit Π for owners (resultingfrom debt devaluation), which for unity capital cor-responds to the difference between relation (6.33a)and (6.32)

Π = (d + r∗)l∗(PK − P0

K

), (6.34)

as clarified by Cohn and Modigliani [6.44].• Subscription of debts at variable interest rate, ableto be adjusted outright to inflation rate to compen-sate completely for debt devaluation, according toFisher’s theory of interest; these possibilities shouldbe rules out, as normally firms are insured againstdebt cost variation since they sign fixed-rate con-tracts.

Finally the only reason supporting the connec-tion (6.33a) remains the first (accounting for inflation),but generally firms’ behavior is intended to calculate cKaccording to relation (6.33b) (accounting to historicalcosts). However, rational behavior should compel theuse of (6.32), thereby avoiding the over- or underesti-mation of capital cost ex ante.

In summary, the prices equation can be writ-ten at a macroeconomics level by substituting (6.32)into (6.30)

P = w/λ+ (d + r∗)v[l∗ P0

K + (1− l∗)PK]. (6.35a)

This relation is simplified according to the differentaims of the enterprises:

1. Keeping physical capital intact, as suggested byaccountancy for inflation, presuming that capitalprice moves in perfect accordance with productprice (P0

K = PK = pk P over pk = PK/P > 1 is thecapital-related price compared with the productone)

Pa = w/λ

1− (d + r∗)vpk= (

1+μ∗a

)w/λ . (6.35b)

2. Integrity of capital conferred by owners, as wouldbe suggested by rational behavior (PK = pk P> P0

K)

Pb = w/λ+ l∗(d + r∗)vP0K

1− (1− l∗)(d + r∗)vpk. (6.35c)

3. Recovery of nominal value of capital, so as only totake account of historical costs (PK = P0

K)

Pc = w/λ+ (d + r∗)vP0K . (6.35d)

Only in the particular case of (6.35b) can the pricelevel be obtained by applying a steady profit-marginfactor (1+μ∗

a ) to labor costs per product unit (w/λ).The markup μ∗ desired by enterprises (the percent-

age calculated on variable work costs in order to recoverfixed capital cost) results in the following expressionsfor the three cases above:

1. Keeping physical capital intact; in this case, themark-up results independent of the nominal wagelevel w

μ∗a = Paλ/w−1 = (d + r∗)vpk

1− (d + r∗)vpk. (6.36a)

2. Integrity of capital conferred by owners,

μ∗b = Pbλ/w−1

= (d + r∗)v[l∗ P0

Kλ/w+ (1− l∗)pk]

1− (1− l∗)(d + r∗)vpk. (6.36b)

3. Recovery of nominal value of capital,

μ∗c = Pcλ/w−1 = (d + r∗)vP0

Kλ/w . (6.36c)

Note that in case 1, enforcing automation gen-erally implies adoption of manufacturing techniqueswhose relative cost pk increases at a rate greater thanproportionally with respect to the capital–product ratereduction v. The desired markup must then be aug-mented in order to ensure coverage of capital. Inrelation (6.35b) this effect is compensated because ofproductivity λ growth due to greater automation, so thaton the whole the effect of automation on the generalprice level is beneficial: for given nominal salary, au-tomation reduces price level.

In cases 2 and 3 of rational behavior, referringto (6.35c) in which the enterprise is aware that its debtis to be refunded at its nominal value, and even more soin the particular case (6.35d) in which the firm is endur-ing monetary illusion, the desired markup is variable,a decreasing function of monetary wage.

Therefore the markup theory is a simplification lim-ited to the case of maintaining physical capital intact,and neglecting effects of capital composition (debt re-fundable at nominal value).

Based on the previous prices equations it followsthat an increase of nominal wages or profit rate soughtby enterprises involves an increase in prices generallevel. In comparison with w, the elasticity is only unityin (6.35b) and diminishes increasingly when passingto (6.35c) and (6.35d).

A percentage increase of nominal salaries is there-fore transferred on the level of prices in the same

PartA

6.4


proportion – as stated by markup theory – only in theparticular case of (6.35b). In the other cases the transla-tion results less than proportional, as the sought markupdecreases as w increases.

Nominal Prices and Wages: Some ConclusionsElasticity of product price decreases if the termvP0

Kλ/w increases, representing the ratio between thenominal value of capital (P0

K K ) and labor cost (wL).Since automation necessarily implies a remarkable in-crease of this ratio, it results in minor prices sensibilityto wages variation. In practice, automation impliesbeneficial effects on inflation: for increasing nominalwages, a high-capital-intensity economy is less subjectto inflation shocks caused by wage rises.

6.4.2 Macroeconomics Effectsof Automation in the Mid-Term:Actual Wagesand Natural Unemployment

In order to analyze effects of automation on macroeco-nomic balance in the mid term, it is necessary to conveythe former equations of prices in terms of real wages,dividing each member by P, in order to take accountof ω = w/P, which represents the maximum wage thatfirms are prepared to pay to employees without givingup their desired profit rate.

1. Keeping physical capital intact (P0K = PK = pk P)

ωa = λ[1− (d + r∗)vpk] = λ/(1+μ∗

a

)(6.37a)

2. Wholeness of capital granted by owners (PK =pk P > P0

K)

ωb = λ{1− (d + r∗)v

[pk − l∗

(pk − P0

K/P)]}

(6.37b)

3. Recovery of capital’s nominal value (PK = P0K)

ωc = λ[1− (d + r∗)vP0

K/P]

(6.37c)

In the borderline case of keeping physical capitalintact (6.37a) only one level of real salary ωa existsconsistent with a specified level of capital productiv-ity r∗, given work productivity λ, amortization rate d,capital/product ratio v (corresponding to normal use ofplants), and relative price pk of capital with respect toproduct.

The value of ωa can also be expressed as a linkbetween work productivity and profit margin factor

(1+μ∗a ) with variable costs, assuming that the desired

markup is unchanging in comparison with prices.In the rational case of corporate stock integrity and

in the generally adopted case of recovery of capitalnominal value, an increasing relation exists betweenreal salary and general price level, with the desiredmarkup being variable, as shown in connections (6.36b)and (6.36c).

The elasticity of ω with respect to P turns out toincrease from (6.37a) to (6.37c)

ηa = (∂ω/∂P)(P/ω) = 0

< ηb = (λ/ωb)l∗(d + r∗)vP0K/P

< ηc = (λ/ωc)(d + r∗)vP0K/P

= λ/ωc −1 <>

1 with ωc><

λ/2 . (6.37d)

In cases 2 and 3 growth in the prices general levelraises the added value share intended for normal cap-ital remuneration, leaving firms a markup to be usedin bargaining in order to grant employees a pay rise inreal salary, even while keeping normal profit rate (thecalculated level with programmed production) steady.

This markup of real wage bargaining depends onthe choice of the capital recovery system: it is higher inthe case of fund illusion and lower in the case in whichcapital stock is intended to be maintained undivided.

Remark: In all three cases above, the level of realsalary depends on the degree of production processautomation: in economic systems where this is high,productivity of work (λ) and capital (1/v) are neces-sarily higher than the related price of capital (pk andP0

K/P), so the real wage that the firms are willingto concede in bargaining is higher than the one af-fecting work-intensive economic systems. In substance,automation weakens firms’ resistance in wage bargain-ing, making them more willing to concede higher realwages.

In order to qualify the positive role of automa-tion we complete the mid-term period macroeconomicmodel with the wage equation from the workers’ side.

Macroeconomics theory believes that, in the midterm, nominal wages w asked for by workers in syn-dicated or individual bargaining (or imposed by firms inorder to select and extract optimal effort in productionprocess) depend on three variables: the unemploymentrate of man power u, other factors absorbed in a syn-thetic variable z, and the expected level of prices P e.

So it is possible to write

w = ω(u, z)P e . (6.38)

PartA

6.4


Higher unemployment is supposed to weaken workers’contractual strength, compelling them to accept lowerwages, and vice versa that a decrease of unemploy-ment rate leads to requests for an increase of real wages(∂ω/∂u < 0).

The variable z can express the effects of unemploy-ment increase, which would compel workers to ask fora pay raise, because any termination would appear lessrisky in terms of social salary (the threshold abovewhich an individual is compelled to work and underwhich he is not prepared to accept, at worst choos-ing unemployment). Similar effects would be inducedby legal imposition of minimum pay and other formsof worker protection, which – making discharge moredifficult – would strengthen workers’ position in wagebargaining.

Regarding the expected level of prices, it is sup-posed that

∂w/∂P e = w/P e > 0 ⇒ (∂w/∂P e)(P e/w) = 1

to point out that rational subjects are interested inreal wages (their buying power) and not nominal ones,so an increase in general level of prices would bringabout an increase in nominal wages in the same pro-portion. Wages are however negotiated by firms onnominal terms, on the basis of a price foresight (P e) forthe whole length of the contract (generally more than1 year), during which monetary wages are not correctedif P �= P e.

To accomplish this analysis, in the mid-term period,wage bargaining is supposed to have real wage as a sub-ject (excluding systematic errors in expected inflationforesight), so that P = P e and (6.38) can be simplifiedto

ω = ω(u, z) . (6.39)

Figure 6.2 illustrates this relation with a decreasingcurve WS (wage setting), on Cartesian coordinates fora given level of z.

Two straight lines, PS (price setting), representingprice equations (6.37a), considering mainly the bor-derline case of maintaining physical capital intact, arereported:

• The upper line PSA refers to an economic sys-tem affected by a high degree of automation, orby technologies where θk = 1 and work and capi-tal productivity are higher, being able to contrast thecapital’s relative higher price; in this case firms areprone to grant higher real wages.

u An ua

n

Ea

EAPSA

PSa

WS(z)

u

ω

0

Fig. 6.2 Mid-term equilibrium with high and low automa-tion level

• The lower line PSa, characterizing an economy witha lower degree of automation, resulting in less will-ingness to grant high real wages.

The intersection between the curve WS and the line PSindicates one equilibrium point E for each economicsystem (where workers’ plans are consistent with thoseof firms), corresponding to the case in which only onenatural unemployment rate exists, obtained by balanc-ing (6.37a) and (6.39)

λ[1− (d + r∗)vpk

] = ω(u, z) ⇒ uAn < ua

n , (6.40a)

assuming that [1 − (d + r∗)vpk]dλ− λ(d + r∗)pkdv >

λ(d + r∗)vdpk.

Remark: In the case of a highly automated system,wage bargaining in the mid-term period enables a natu-ral unemployment rate smaller than that affecting a lessautomated economy: higher productivity makes enter-prises more willing to grant higher real wages as a resultof bargaining and the search for efficiency in produc-tion organizations, so that the economic system canconverge towards a lower equilibrium rate of unemploy-ment (uA

n < uan).

The uniqueness of this solution is valid only if en-terprises aim to maintain physical capital undivided.

In the case of rational behavior intended to achievethe integrity of capital stock alone (6.37b), or businessprocedures oriented to recoup the accounting value ofcapital alone (6.37c), the natural unemployment ratetheory is not sustainable. In fact, if we equalize thesetwo relations to (6.39), respectively, we obtain in bothcases a relation describing balance couples between un-employment rate and prices general level, indicatingthat the equilibrium rate is undetermined and therefore

PartA

6.4


that the adjective natural is inappropriate

λ{1− (d + r∗)v

[pk − l∗

(pk − P0

K/P)]} = ω(u, z)

⇒ un = ubn(P) , (6.40b)

λ[1− (d + r∗)vP0

K/P] = ω(u, z)

⇒ un = ucn(P) , (6.40c)

where ∂un/∂P < 0, since an increase in P increasesthe margin necessary to recover capital cost and soenterprises can pay out workers a higher real salary(∂ω/∂P > 0), which in balance is compatible witha lower rate of unemployment.

Summing up, alternative enterprise behavior to thatintended to maintain physical capital intact does notallow fixing univocally the natural rate of unemploy-ment. In fact, highly automated economic systems causea translation of the function un = un(P) to a lowerlevel: for an identical general level of prices the unem-ployment balance rate is lower since automation allowsincreased productivity and real wage that enterprises areprepared to pay.

6.4.3 Macroeconomic Effectsof Automation in the Mid Term:Natural Unemploymentand Technological Unemployment

The above optimistic conclusion of mid-term equi-librium being more profitable for highly automationeconomies must be validated in the light of constraintsimposed by capital-intensive technologies, where theproduction rate is higher and yields are constant.

Assume that in the economic system an aggregateddemand is guaranteed, such that all production volumescould be sold, either owing to fiscal and monetary poli-tics oriented towards full employment or hoping that, inthe mid term, the economy will spontaneously convergethrough monetary adjustments induced by variation ofthe price general level.

In a diffused automation context, a problem couldresult from a potential inconsistency between the unem-ployment natural rate un (obtained by imposing equalityof expected and actual prices, related to labor marketequilibrium) and the unemployment rate imposed bytechnology u

u = 1− Ld/Ls = 1− (Y/λ)/Ls ><

un , (6.41)

where Ld = Y/λ is the demand for labor, Ls is the laborsupply, and Y is the potential production rate, compati-ble with full utilization of production capacity.

A justification of this conclusion can be seen bynoting that automation calls for qualified technical per-sonnel, who are still engaged by enterprises even duringcrisis periods, and of whom overtime work is requiredin case of expansion. Then, employment variation isnot proportional to production variation, as clearly re-sults from the law of Okun [6.45], and from papers byPerry [6.46] and Tatom [6.47].

Remark: From the definition of u it follows thateconomic systems with high automation level, andtherefore characterized by high productivity of labor,present higher technological unemployment rates

(∂u/∂λ)(λ/u) = (1− u)/u > 0 .

On one hand, automation reduces the natural unemploy-ment rate un, but on the other it forces the technologicalunemployment rate u to increase. If it holds that u ≤ un,the market is dominant and the economic system aims toconverge in time towards an equilibrium characterizedby higher real salary and lower (natural) unemploy-ment rate, as soon as the beneficial effects of automationspread.

In Fig. 6.3, the previous Fig. 6.2 is modified un-der the hypothesis of a capital-intensive economy, byincluding two vertical lines corresponding to two poten-tial unemployment rates imposed by technology (u anduA). Note that u has been placed on the left of un whilstuA has been placed on the right of un. It can be seenthat the labor market equilibrium (point E) dominateswhen technology implies a degree of automation com-patible with a nonconstraining unemployment rate u.Equilibrium could be obtained, on the contrary, whentechnology (for a given production capacity of the eco-

uAu un

E

H

PSA

WS(z)

u

ω

0

Fig. 6.3 Mid-term equilibrium with high automation leveland two different technological unemployment rates

PartA

6.4


nomic system) cannot assure a low unemployment rate:uA > un.

The technological unemployment constraint impliesa large weakness when wages are negotiated by work-ers, and it induces a constrained equilibrium (point H)in which the real salary perceived by workers is lowerthan that which enterprises are willing to pay. The lattercan therefore achieve unplanned extra profits, so that au-tomation benefits only generate capital owners’ incomewhich, in the share market, gives rise to share valueincrease.

Only a relevant production increase and equivalentaggregated demand could guarantee a reduction of uA,thus transferring automation benefits also to workers.

These conclusions can have a significant im-pact on the Fisher–Phillips curve (Fisher [6.48]and Phillips [6.49], first; theoretically supported byLipsey [6.50] and then extended by Samuelson andSolow [6.51]) describing the trade-off between inflationand unemployment.

Assume that the quality constraint between ex-pected prices and effective prices can, in the mid term,be neglected, in order to analyze the inflation dynamics.Then, substitute (6.38) into (6.35b), by assuming the hy-pothesis that maintaining capital intact implies constantmarkup

P = (1+μ∗)ω(u, z)P e/λ . (6.42)

This relation shows that labor market equilibrium forimperfect information (P e �= P), implies a positive linkbetween real and expected prices.

By dividing (6.42) by P−1, the following relationresults

1+π = (1+π e)(1+μ∗)ω(u, z)/λ , (6.43a)

where:

• π = P/P−1 −1 is the current inflation rate• π e = P e/P−1 −1 is the expected inflation rate.

Assume moderate inflation rates, such that

(1+π)/(1+π e) ≈ 1+π −π e

and consider a linear relation between productivity andwages, such that it amounts to 100% if

z = u = 0 : ω(u, z)/λ = 1+α0z −α1u(α0, α1 > 0) .

Relation (6.43a) can be simplified to

π = π e + (1+μ∗)α0z − (1+μ∗)α1u , (6.43b)

which is a linear version of the Phillips curve; accordingto this form, the current inflation rate depends positively

on inflation expectation, markup level, and the variablez, and negatively on the unemployment rate.

For π e = 0, the original curve which Phillips andSamuelson–Solow estimated for the UK and USA isobtained.

For π e = π−1 (i. e., extrapolative expectations)a link between the variation of inflation rate and theunemployment rate (accelerated Phillips curve), show-ing better interpolation of data observed since 1980s, isderived

π −π−1 = (1+μ∗)α0z − (1+μ∗)α1u . (6.44)

Assuming π = π e = π−1 in (6.43b) and (6.44), anestimate of the natural unemployment rate (without sys-tematic errors in mid-term forecasting) can be derived

un = α0z/α1 . (6.45)

Now, multiplying and dividing by the α1 term (1 +μ∗)α0z in (6.44), and inserting (6.45) into (6.44), it re-sults that, in the case of extrapolative expectations, theinflation rate reduces if the effective unemployment rateis greater than the natural one (dπ < 0 if u > un); it in-creases in the opposite case (dπ > 0 if u < un); and itis zero (constant inflation rate) if the effective unem-ployment rate is equal to the natural one (dπ = 0 ifu = un)

π −π−1 = −(1+μ∗)α1(u −un) . (6.46)

Remark: Relation (6.44) does not take into accounteffects induced by automation diffusion, which imposeson the economic system a technological unemploymentuA that increases with increasing automation.

It follows that any econometric estimation basedon (6.44) no longer evaluates the natural unemploymentrate (6.45) for π −π−1 = 0, because the latter varies intime, flattening the interpolating line (increasingly highunemployment rates related to increasingly lower realwages, as shown above). Then, as automation processspreads, the natural unemployment rate (which, with-out systematic errors, assures compatibility between thereal salary paid by enterprises and the real salary ei-ther demanded by workers or supplied by enterprisesfor efficiency motivation) is no longer significant.

In Fig. 6.4a the Phillips curve for the Italian eco-nomic system, modified by considering inflation ratevariations from 1953 to 2005 and the unemploymentrate, is reported; the interpolation line is decreasing butvery flat owing to uA movement towards the right.

A natural unemployment rate between 7 and 8%seems to appear, but the intersection of the interpola-tion line with the abscissa is moved, as shown in the

PartA

6.4


4.5 5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10 10.5 11 11.5 12

Inflation rate variation (%)

Unemployment rate (%)

876543210

–1–2–3–4–5–6

4.8 5.2 5.6 6

Inflation rate variation (%)a) Economic miracle: 1953–1972


8

6

4

2

0

–2

–4

–66 7 8 9 10

Inflation rate variation (%)b) Petroleum shock: 1973– 1985


8

6

4

2

0

–2

–4

–66 7 8 9 10 11 12

Inflation rate variation (%)c) 1973–2005


8

6

4

2

0

–2

–4

–6

Fig. 6.4a–c Italy: Expectations-augmented Phillips curve (1953–2005). (a) Economic miracle: 1953–1972,(b) petroleum shock: 1973–1985, (c) 1973–2005

three representations where homogeneous data sets ofthe Italian economic evolution are considered.

Automation has a first significant impact during theoil crisis, moving the intersection from 5.2% for thefirst post-war 20 years up to 8% in the next period. Ex-tending the time period up to 2005, intersection movesto 9%; the value could be greater, but it has been re-cently bounded by the introduction of temporary workcontracts which opened new labor opportunities butlowered salaries.

Note that Figs. 6.4a–c are partial reproductions ofthe Fig. 6.4, considering three different intervals of un-employment rate values. This reorganization of datacorresponds to the three different periods of the Italianeconomic growth: (a) the economic miracle (1953–

1972), (b) the petroleum shock (1973–1985), and (c) theperiod from the economic miracle until 2005 (1973–2005).

A more limited but comprehensive analysis, ow-ing to lack of homogeneous historical data over thewhole period, has been done also for two other impor-tant European countries, namely France (Fig. 6.5a) andGermany (Fig. 6.5b): the period considered ranges from1977 to 2007.

The unemployment rate values have been recom-puted so as to make all the historical series homoge-neous. Owing to the limited availability of data for bothcountries, only two periods have been analyzed: theperiod 1977–1985, with the effects of the petroleumshock, and the whole period up to 2007, in order to

PartA

6.4


4 4.5 5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10 10.5 11 11.5 12

a) Inflation rate variation (%) France: Expectations-augmented Phillips curve (1977–2007)


32.5

21.5

10.5

0–0.5

–1–1.5

–2–2.5

–3–3.5

–4

4 4.5 5 5.5 6 6.5 7 7.5 8 8.5 9 9.5

Inflation rate variation (%) France: Petroleum shock: 1977–1985


32.5

21.5

10.5

0–0.5

–1–1.5

–2–2.5

–3–3.5

–4

Fig. 6.5 (a) France: expectations-augmented Phillips curve (1977–2007) (b) � Germany: expectations-augmentedPhillips curve (1977–2007)

evaluate the increase of the natural unemployment ratefrom the intersection of the interpolation line with theabscissa.

As far as Fig. 6.5a is concerned, the natural unem-ployment rate is also increased in France – as in Italy –by more than one percentage point (from less than 6%to more than 7%). In the authors’ opinion, this increaseshould be due to technology automation diffusion andconsequent innovation of organization structures.

Referring to Fig. 6.5b, it can be noted that even inGermany the natural unemployment rate increased –more than in France and in Italy – by about three per-centage points (from 2.5 to 5.5%). This effect is partlydue to application of automated technologies (which

should explain about one percentage point, as in theother considered countries), and partly to the reunifica-tion of the two parts of German.

A Final Remark: Based on the considerations anddata analysis above it follows that the natural unem-ployment rate, in an industrial system where capitalintensive enterprises are largely diffused, is no longersignificant, because enterprises do not apply methodsto maintain capital intact, and because technologicalinflation tends to constraint the natural one.

Only structural opposing factors could slow thisprocess, such as labor market reform, to give rise tonew work opportunities, and mainly economic poli-tics, which could increase the economy development

PartA

6.4

Economic Aspects of Automation 6.5 Final Comments 111

0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10 10.5 11

b) Inflation rate variation (%) Germany: Expectations-augmented Phillips curve (1977–2007)


2

1.5

1

0.5

0

–0.5

–1

–1.5

–2

–2.5

–3

2

1.5

1

0.5

0

–0.5

–1

–1.5

–2

–2.5

–3

0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6

Inflation rate variation (%) Germany: Petroleum shock: 1977–1985


trend more than the growth of productivity and laborsupply. However, these aspects should be approached

in a long-term analysis, and exceed the scope of thischapter.

6.5 Final Comments

Industrial automation is characterized by dominanceof capital dynamics over the biological dynamics ofhuman labor, thus increasing the production rate. There-fore automation plays a positive role in reducing costsrelated to both labor control, sometimes making mon-etary incentives useless, and supervisors employed toassure the highest possible utilization of personnel. It isautomation itself that imposes the production rate andforces workers to correspond to this rate.

In spite of these positive effects on costs, the in-creased capital intensity implies greater rigidity ofthe cost structure: a higher capital cost depending onhigher investment value, with consequent transforma-tion of some variable costs into fixed costs. This inducesa greater variance of profit in relation to productionvolumes and a higher risk of automation in respect tolabor-intensive systems. However, the trade-off betweenautomation yield and risk suggests that enterprises

PartA

6.5


1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

Computerrelated services

Total business sector

Total ICT sector

ICT services

Business services

Employment growth (%)60

50

40

30

20

10

0

Fig. 6.6 Employment growth by sector with special focus on computer and ICT (source: OECD Information TechnologyOutlook 2008, based on STAN database)

should increase automation in order to obtain high uti-lization of production capacity. One positive effort inthis direction has been the design and implementationof flexible automation during recent decades (e.g., seeChap. 50 on Flexible and Precision Assembly).

Moving from microeconomic to macroeconomicconsiderations, automation should reduce the effectson short-term inflation caused by nominal salary in-creases since it forces productivity to augment morethan the markup necessary to cover higher fixed costs.In addition, the impact of automation depends on thebook-keeping method adopted by the enterprise in orderto recover the capital value: this impact is lower if a partof capital can be financed by debts and if the enterprisebehavior is motivated by monetary illusion.

Pola

nd

Tur

key

EU

15

Aus

tral

ia

Can

ada

Uni

ted

Stat

es

Lux

embo

urg

Uni

ted

Kin

gdom

Den

mar

k

Finl

and

Swed

en

Net

herl

ands

Ital

y

Bel

gium

Ger

man

y

Irel

and

Aus

tria

Fran

ce

Spai

n

Gre

ece

Port

ugal

Nor

way

Switz

erla

nd

Hun

gary

Icel

and

Cze

ch R

epub

lic

Slov

enia

Est

onia

Slov

ak R

epub

lic

Share (%)

1995 200735

30

25

20

15

10

5

0

Fig. 6.7 Share of ICT-related occupations in the total economy from 1995 and 2007 (source: OECD IT Outlook 2008,forthcoming)

Over a mid-term period, automation has been rec-ognized to have beneficial effects (Figs. 6.6–6.8) bothon the real salary paid to workers and on the (natural)unemployment trend, if characteristics of automationtechnologies do not form an obstacle to the market trendtowards equilibrium, i. e., negotiation between enter-prises and trade unions. If, on the contrary, the systemproduction capacity, for a compatible demand, preventsmarket convergence, a noncontractual equilibrium onlydependent on technology capability could be estab-lished, to the advantage of enterprises and the prejudiceof workers, thus reducing real wages and increasing theunemployment rate.

Empirical validation seems to show that Italyentered this technology-caused unemployment phase

PartA

6.5

Economic Aspects of Automation 6.6 Capital/Labor and Capital/Product Ratios in the Most Important Italian Industrial Sectors 113

Aus

tral

ia

Uni

ted

Stat

es

Swed

en

Den

mar

k

Uni

ted

Kin

gdom

Bel

gium

Can

ada

Japa

n

New

Zea

land

Spai

n

Net

herl

ands

Port

ugal

Finl

and

Irel

and

Gre

ece

Ital

y

Ger

man

y

Fran

ce

Aus

tria

%

1990–95 1995–20031

0.90.80.70.60.50.40.30.20.1

0

Fig. 6.8 Contributions of ICT investment to GDP growth, 1990–1995 and 1995–2003 in percentage points (source:OECD Productivity Database, September 2005)

during the last 20 years: this corresponds to a flatterPhillips curve because absence of inflation acceler-ation is related to natural unemployment rates that

increase with automation diffusion, even if some struc-tural modification of the labor market restrained thistrend.

6.6 Capital/Labor and Capital/Product Ratiosin the Most Important Italian Industrial Sectors

A list of the most important industrial sectors inItaly is reported in Table 6.2 (data collected byMediobanca [6.52] for the year 2005). The followingestimations of the model variables and rates have beenadopted (as enforced by the available data): capital Kis estimated through fixed assets; with reference to pro-duction, the added value Va is considered; labor L isestimated in terms of number of workers.

Some interesting comments can be made based onTable 6.2, by using the production function models pre-sented in the previous sections.

According to the authors’ experience (based on theirknowledge of the Italian industrial sectors), the follow-ing anomalous sectors can be considered:

• Sectors concerning energy production and distribu-tion and public services for the following reasons:they consist of activities applying very high levelsof automation; personnel applied in production areextremely scarce, whilst it is involved in organiza-tion and control; therefore anomalous values of thecapital/labor ratio and of productivity result.

• The transport sector, because the very high capi-tal/product ratio depends on the low level of theadded value of enterprises that are operating withpolitically fixed (low) prices.

0 1 2 3 4 5 6

Capital/labor (×1000 Euro)

Capital/production (rate)

600

500

400

300

200

100

0

Fig. 6.9 Correlation between capital/labor (× 1000 €) andcapital/production (ratios)

PartA

6.6


Table 6.2 Most important industrial sectors in Italy (after [6.52])

Industrial sector Fixed assets Added Number Capital/ Capital/ Productivity

(year 2005) (million value of workers product labor (× 1000

euro) (a) (million (× 1000) (a/b) (× 1000 euro) (b/c)

euro) (b) (c) euro) (a/c)

Industrial enterprises 309 689.6 84 373.4 933.8 3.7 331.6 90.4

Service enterprises 239 941.9 42 278.3 402.7 5.7 595.8 105.0

Clothing industry 2824.5 1882.9 30.4 1.5 93.0 62.0

Food industry: drink production 5041.7 1356.2 14.3 3.7 352.1 94.7

Food industry: milk & related products 1794.6 933.2 10.1 1.9 177.8 92.5

Food industry: alimentary preservation 2842.4 897.8 11.2 3.2 252.9 79.9

Food industry: confectionary 4316.1 1659.0 17.8 2.6 242.7 93.3

Food industry: others 5772.7 2070.2 23.9 2.8 241.4 86.6

Paper production 7730.9 1470.6 19.9 5.3 388.2 73.9

Chemical sector 16 117.7 4175.2 47.7 3.9 337.7 87.5

Transport means production 21 771.4 6353.3 121.8 3.4 178.8 52.2

Retail distribution 10 141.1 3639.8 84.2 2.8 120.4 43.2

Electrical household appliances 4534.7 1697.4 35.4 2.7 128.1 47.9

Electronic sector 9498.9 4845.4 65.4 2.0 145.1 74.0

Energy production/distribution 147 200.4 22 916.2 82.4 6.4 1786.8 278.2

Pharmaceuticals & cosmetics 9058.9 6085.2 56.3 1.5 160.8 108.0

Chemical fibers 2081.4 233.2 5.0 8.9 418.4 46.9

Rubber & cables 3868.3 1249.7 21.2 3.1 182.1 58.8

Graphic & editorial 2399.1 1960.4 18.0 1.2 133.3 108.9

Plant installation 1295.9 1683.4 23.2 0.8 56.0 72.7

Building enterprises 1337.7 1380.0 24.7 1.0 54.2 55.9

Wood and furniture 2056.1 689.2 12.8 3.0 160.2 53.7

Mechanical sector 18 114.1 9356.6 137.3 1.9 131.9 68.1

Hide and leather articles 1013.9 696.6 9.0 1.5 113.1 77.7

Products for building industry 11 585.2 2474.3 28.7 4.7 404.3 86.4

Public services 103 746.7 28 413.0 130.5 3.7 795.2 217.8

Metallurgy 18 885.6 5078.2 62.4 3.7 302.9 81.4

Textile 3539.5 1072.1 21.7 3.3 163.2 49.4

Transport 122 989.3 7770.5 142.1 15.8 865.2 54.7

Glass 2929.3 726.2 9.2 4.0 317.8 78.8

Notation: Labor-intensive sectors Intermediate Anomalous

• The chemical fibres sector, because at the time con-sidered it was suffering a deep crisis with a largeamount of unused production capacity, which gaverise to an anomalous (too high) capital/product ratioand anomalous (too small) value of productivity.

Sectors with a rate capital/production of 1.5 (suchas the clothing industry, pharmaceutics and cosmet-ics, and hide and leather articles) has been consideredas intermediate sectors, because automation is highlyimportant in some working phases, whereas other work-

ing phases still largely utilize workers. These sectors(and the value 1.5 of the rate capital/production) areused as separators between capital-intensive systems(with high degrees of automation) and labor-intensiveones.

Sectors with a capital/production ratio of less than1.5 have been considered as labor-intensive systems.Among these, note that the graphic sector is capital in-tensive, but available data for this sector are combinedwith the editorial sector, which in turn is largely laborintensive.

PartA

6.6

Economic Aspects of Automation References 115

All other sectors can be viewed as capital-intensivesectors, even though it cannot be excluded that someworking phases within their enterprises are still laborintensive.

Two potential correlations, if any, are illustrated inthe next two figures: Fig. 6.9 shows the relations be-tween the capital/labor and capital/production ratios,and Fig. 6.10 shows the relations between productivityand capital/labor ratio.

As shown in Fig. 6.9, the capital/labor ratio exhibitsa clear positive correlation with the capital/productionratio; therefore they could be considered as alternativemeasures of capital intensity.

On the contrary, Fig. 6.10 shows that productiv-ity does not present a clear correlation with capital.This could be motivated by the effects of other fac-tors, including the utilization rate of production capacityand the nonuniform flexibility of the workforce, whichcould have effects on productivity.

0 100 200 300 400 500 600

Productivity

Capital/labor

110

100

90

80

70

60

50

40

Fig. 6.10 Relation between productivity (× 100 €) and cap-ital/labor (rate)

References

6.1 CODESNET: Coordination Action No. IST-2002-506673/ Joint Call IST-NMP-1, A. Villa, coordinator. Web siteaddress: www.codesnet.polito.it. (2004–2008)

6.2 R.W. Shephard: Cost and Production Functions(Princeton Univ. Press, Princeton 1953)

6.3 R.W. Shephard: Theory of Cost and Production Func-tions (Princeton Univ. Press, Princeton 1970)

6.4 R. Frisch: Lois Techniques et Economiques de la Pro-duction (Dunod, Paris 1963), (in French)

6.5 H. Uzawa: Duality principles in the theory of costand production, Int. Econ. Rev. 5, 216–220 (1964)

6.6 M. Fuss, D. Mc Fadden: Production Economics: A DualApproach to Theory and Application (North-Holland,Amsterdam 1978)

6.7 C.W. Cobb, P.H. Douglas: A theory of production, Am.Econ. Rev. 18, 139–165 (1928)

6.8 J.K. Arrow, H.B. Chenery, B.S. Minhas, R. Solow:Capital-labor substitution and economic efficiency,Rev. Econ. Stat. 63, 225–47 (1961)

6.9 M. Brown, J.S. De Cani: Technological change andthe distribution of income, Int. Econ. Rev. 4, 289–95(1963)

6.10 D. Mc Fadden: Further results on CES productionfunctions, Rev. Econ. Stud. 30, 73–83 (1963)

6.11 H. Uzawa: Production functions with constant elas-ticity of substitution, Rev. Econ. Stud. 29, 291–99(1962)

6.12 G.H. Hildebrand, T.C. Liu: Manufacturing Produc-tion Functions in the United States (State School ofIndustrial Labor Relations, New York 1965)

6.13 L.R. Christensen, D.W. Jorgenson, L.J. Lau: Conjugateduality and the transcendental logarithmic produc-tion function, Econometrica 39, 255–56 (1971)

6.14 L.R. Christensen, D.W. Jorgenson, L.J. Lau: Transcen-dental logarithmic production frontier, Rev. Econ.Stat. 55, 28–45 (1973)

6.15 I.M. Nadiri: Producers theory. In: Handbook ofMathematical Economics, Vol. II, ed. by K.J. Ar-row, M.D. Intriligator (North-Holland, Amsterdam1982)

6.16 A. Shaik: Laws of production and laws of algebra:the humbug production function, Rev. Econ. Stat.56, 115–20 (1974)

6.17 H. Hotelling: Edgeworth’s taxation paradox and thenature of demand and supply functions, J. Polit.Econ. 40, 577–616 (1932)

6.18 H. Hotelling: Demand functions with limited bud-gets, Econometrica 3, 66–78 (1935)

6.19 P.A. Samuelson: Foundations of Economic Analysis(Harvard Univ. Press, Cambridge 1947)

6.20 P.A. Samuelson: Price of factors and goods in generalequilibrium, Rev. Econ. Stud. 21, 1–20 (1954)

6.21 W.E. Diewert: Duality approaches to microeconomictheory. In: Handbook of Mathematical Economics,Vol. II, ed. by K.J. Arrow, M.D. Intriligator (North-Holland, Amsterdam 1982)

6.22 D.W. Jorgenson: Econometric methods for modellingproducer behaviour. In: Handbook of Econometrics,Vol. III, ed. by Z. Griliches, M.D. Intriligator (North-Holland, Amsterdam 1986)

PartA

6

www.codesnet.polito.it


6.23 E. Luciano, P. Ravazzi: I Costi nell’Impresa. Teo-ria Economica e Gestione Aziendale (UTET, Torino1997), (Costs in the Enterprise. Economic Theory andIndustrial Management – in Italian)

6.24 A. Weiss: Efficiency Wages (Princeton Univ. Press,Princeton 1990)

6.25 J.E. Stiglitz: Wage determination and unemploy-ment in LDC’s: the labor turnover model, Q. J. Econ.88(2), 194–227 (1974)

6.26 S. Salop: A model of the natural rate of unemploy-ment, Am. Econ. Rev. 69(2), 117–25 (1979)

6.27 A. Weiss: Job queues and layoffs in labor mar-kets with flexible wages, J. Polit. Econ. 88, 526–38(1980)

6.28 C. Shapiro, J.E. Stiglitz: Equilibrium unemploymentas a worker discipline device, Am. Econ. Rev. 74(3),433–44 (1984)

6.29 G.A. Calvo: The inefficiency of unemployment: thesupervision perspective, Q. J. Econ. 100(2), 373–87(1985)

6.30 G.A. Akerlof: Labor contracts as partial gift exchange,Q. J. Econ. 97(4), 543–569 (1982)

6.31 G.A. Akerlof: Gift exchange and efficiency-wagetheory: four views, Am. Econ. Rev. 74(2), 79–83(1984)

6.32 H. Miyazaki: Work, norms and involontary unem-ployment, Q. J. Econ. 99(2), 297–311 (1984)

6.33 D. Antonelli, N. Pasquino, A. Villa: Mass-customizedproduction in a SME network, IFIP Int. Working Conf.APMS 2007 (Linkoping, 2007)

6.34 J. Robinson: The Economics of Imperfect Competition(Macmillan, London 1933)

6.35 E.H. Chamberlin: The Theory of Monopolistic Compe-tition (Harvard Univ. Press, Harvard 1933)

6.36 P.W.S. Andrews: On Competition in Economic Theory(Macmillan, London 1964)

6.37 F. Modigliani, M. Miller: The cost of capital, corpo-ration finance and the theory of investment, Am.Econ. Rev. 48(3), 261–297 (1958)

6.38 F. Modigliani, M. Miller: Corporate income taxes andthe cost of capital: a correction, Am. Econ. Rev. 53,433–443 (1963)

6.39 D.N. Baxter: Leverage, risk of ruin and the cost ofcapital, J. Finance 22(3), 395–403 (1967)

6.40 K.H. Chen, E.H. Kim: Theories of corporate debt pol-icy: a synthesis, J. Finance 34(2), 371–384 (1979)

6.41 M.F. Hellwig: Bankruptcy, limited liability and theModigliani–Miller theorem, Am. Econ. Rev. 71(1),155–170 (1981)

6.42 M. Jensen, W. Meckling: Theory of the firm: man-agerial behaviour, agency costs and ownershipstructure, J. Financial Econ. 3(4), 305–360 (1976)

6.43 A.C. Pigou: Maintaining capital intact, Economica45, 235–248 (1935)

6.44 R.A. Cohn, F. Modigliani: Inflation, rational valua-tion and the market, Financial Anal. J. 35, 24–44(1979)

6.45 A.M. Okun: Potential GNP: its measurement and sig-nificance. In: The Political Economy of Prosperity,ed. by A.M. Okun (Brookings Institution, Washington1970) pp. 132–145

6.46 G. Perry: Potential output and productivity, Brook.Pap. Econ. Activ. 8, 11–60 (1977)

6.47 J.A. Tatom: Economic Growth and Unemployment:A Reappraisal of the Conventional View (FederalReserve Bank of St. Louis Review, St. Louis 1978)pp. 16–22

6.48 I. Fisher: A statistical relation between unemploy-ment and prices changes, Int. Labour Rev. 13(6)785–792 (1926), J. Polit. Econ. 81(2), 596–602 (1973)

6.49 W.H. Phillips: The relation between unemploymentand the rate of change of money wages rated inthe United Kingdom: 1861–1957, Economica 25(100),283–299 (1958)

6.50 R.G. Lipsey: The relation between unemploymentand the rate of change of money wages in UK:1862-1957, Economica 27, 1–32 (1960)

6.51 P.A. Samuelson, R.M. Solow: The problem of achiev-ing and maintaining a stable price level: analyticalaspects of anti-inflation policy, Am. Econ. Rev. 50,177–194 (1960)

6.52 Mediobanca: Dati Cumulativi di 2010 Società Italiane(Mediobanca, Milano 2006), (Cumulative Data of 2010Italian Enterprises – in Italian)

PartA

6

Documents

Economic As 6. Economic Aspects of Automation pextras.springer.com/2009/978-3-540-78830-0/... · Economic As 6. Economic Aspects of Automation p Piercarlo Ravazzi, Agostino Villa