Question 1

Generate summary statistics for output (measured by turnover), capital (measured by assets) and labour (measured by employees). Report and briefly describe
the results.

Note: Turnover and Assets are in Euros 000s. Employees are in numbers.
Comment that average (mean) output (measured by Turnover) is €185.68 million, capital (measured by Assets) is €285.23 million and employees are 475. In
addition, there is significant variation between firms in the sample on which they may comment. For instance, turnover ranges from €41.785 billion for the largest
firm by turnover to €2212.6. Assets vary from €126.730 billion to €8819.3 and employee numbers from 106 090 to 1.
We may also comment on the skewness of the data, particularly the difference between mean and median values of turnover and assets, highlighting the positive
skew of the distribution. Scientific notation: 1.33e+006. 1.33 is coefficient. e is 10 to the power of. 6 is exponent. 1.33e+006 = 1.33 x 10 to power of 6 = 1,330,000
 2. How to generate Summary Statistics with gretl. Statistics reported.
On the Toolbar, click on ‘View’.
On the Menu, click on ‘Summary Statistics’.
Enter one of the three variables by highlighting it, then clicking on the rightwards arrow.
Repeat, until all three variables appear in the righthandside column.
Enter.
Definitions. ‘Std. Dev.’ is standard deviation. ‘C.V.’ is coefficient of variation: a statistical measure of the
dispersion of data points in a data series around the mean.
C.V. = Standard Deviation / Mean
In finance, if the C.V. results in a lower ratio of standard deviation to mean return, then the lower (better) is
the risk-return trade-off.
‘Skewness’ is a measure of a dataset’ symmetry. A perfectly symmetrical data set has skewness = 0. Positive
skewness means the right-hand tail will be longer than the left-hand tail. Kurtosis is a measure of the
combined weight of the tails relative to the rest of the distribution. Statistics for skewness and kurtosis do not
give useful information not already given by dispersion?
95th percentile is the highest value left when the top 5% of a numerically sorted data set is discarded.
IQ range is the difference between the 75th and 25th percentiles of data.
How to generate Scatterplot. Click on ‘View’. Choose ‘Graph specified variables’. Choose ‘X-Y Scatterplot’.
Enter your chosen variables.
3. Question 2
Using scatter plots and correlation coefficients, assess the strength of the linear association between log output
and each of the two inputs, log labour and log capital. (

.
4. From the correlation coefficient and scatterplot above, we can answer Question 3.

The correlation coefficient for l_output and l_capital is 0.88 which is strong evidence of a positive linear correlation.
A perfect positive linear correlation is 1.
The scatterplot shows that most of the data points approximate a linear fit, so a positive linear correlation seems to
be determined by the majority of data points across the whole distribution. There is some evidence of outlier firms
with low turnover against asset values, but these make up a relatively small number of data points compared to the
overall sample, and are distributed across the range of values of l_assets.

Lets define a number as N. The natural logarithm of N is (ln N). The natural log of N is the power or exponent by
which the Napier constant, e, has to be raised in order that the result equals N. The Napier constant is approx.
2.718282.

Regarding TMA02 Part B, the 29th observation of Employees is 1 = N, when observations are placed in numerical
order.
(ln 1) = 0; 2.718282 to the power 0 = 1 = N.
The 30th observation of Employees is 2 = N, when observations are placed in numerical order.
(ln 2) = 0.6932; 2.718282 to the power 0.6932 = 2 = N.

One of the basic properties of logs is the Product rule: (ln (a x b)) = ln(a) + ln(b).
This will be useful as we come to construct econometric models.
The Power rule is, ln (a to the power of b) = b ln(a). Any log transformation can be used to transform a model that is
non-linear in parameters into one with desired linearity (linearity in parameters is one OLS assumption.
5. Question 3
Write the linear econometric model that represents a production function in which log of output is a
linear function of log of capital. Estimate the coefficients using a simple (univariate) regression OLS
model. Explain your results.
The model can be written as: l_turnover = a + b x l_assets + u.
(Where a and b are coefficients to be estimated and u is the error term.)
Note: We can use other names for the variables, such as log(output) and log(capital).
So long as the meaning is clear it is fine.

The results indicate that l_turnover is positively related to the log of capital
because the coefficient of l_assets is positive: b = 0.87. The slope coefficient of
l_assets, b, is an estimate of the output elasticity of capital and indicates that
a 1% change in assets leads to a 0.87% change in turnover. Alternatively,
capital elasticity of output, matches the logic of price elasticity of demand.
 6. Question 4
Extend the model of the production function that you wrote in your answer to
. Question 3, by
adding log of labour into your equation and write down the extended (multivariate)
model equation.
Estimate its coefficients and explain the results using elasticities.
l_turnover = a + b x l_assets + c x l_employees + u .
(Where a, b and c are coefficients to be estimated and u is the error term.)
The results are indicated below
7

Results show that l_turnover is positively related to both the log of capital (l_assets) and the
log of labour (l_employees). In the multivariate model that includes a labour variable, the
coefficient of l_assets, b, (the estimate of the output elasticity of capital) is 0.67, which means
that a 1% change in assets leads to a 0.67% change in turnover.
We can also see that the coefficient of log labour, c, which is an estimate of the
output elasticity of labour, is 0.35, which means that a 1% change in the number of employees
leads to a 0.35% change in turnover.
8. Question 5
 Comment on the estimates of the output elasticity of capital
obtained, comparing the univariate and multivariate
regression models in Question 3 and Question 4.
Note that the slope coefficient estimated for the univariate
model of Question 3 estimated a total elasticity encompassing
all effects on turnover, those explicitly included such as log
assets, and those not included, such as log employees.
 By including the log of labour in the model, the estimate of
the output elasticity of capital (the coefficient of l_assets, b)
becomes 0.67, which is lower than in the case of the
univariate model, where it was 0.87. The difference between
the two estimates is due to the inclusion of the labour variable
in the multivariate model. In the multivariate model, the
coefficient of l­_­assets is an estimate of the partial effect of
l_assets on l_turnover, whereas in the univariate model it
represented the total effect since the labour variable was not
included in the model.
 
 Market power and market structure. Chapter 7,
Section 2.2
Competition is the drive to get ahead of rivals. It offers good results for
society, (1) cuts price, (2) offers more choice, (3) produces innovation.
In practice, competition policy cares about two main interrelated factors:
the exercise of market power, and market structure: e.g. barriers to entry.
Competition policy addresses itself to maintaining sufficient levels of
competition in markets. Textbook, p.306 cites EC interventions of (1)
merger leading to market dominance (Ryanair’s bid to acquire Aer Lingus,
main competitor on flights in and out of Dublin airport), (2) conspiring to
set high prices in vitamins A and E markets (Hoffman-La Roche and BASF),
(3) market entry restriction (Microsoft including Media Player within the
Windows operating system, thereby undermining demand for another
music/video management application.
A firm’s market power can be defined as its ability to raise the price of its
output above its marginal cost.
The problem is that prices that are set above marginal cost entail a loss of
welfare for society as a whole.
The focus is on the market for products of a specific industry, not on firm.
 Competition and welfare.
An industry is a set of producers of similar goods or services that compete
directly with each other in a market.
p.311. The market power of each firm is itself hard to measure. In practice
most companies sell a variety of products, and some production is carried
out jointly, by sharing equipment and manpower. The products of a firm
can belong to different industries.
p.308. Consumer surplus is the benefit to consumers, in money terms, of
a specific quantity of a good. It is the difference between the maximum
price consumers would have been willing to pay for each additional unit
of the good (shown by the demand curve) and the purchase price.
p.308. Producer surplus is the difference between the selling price of a
specific quantity of a good and the minimum price at which producers
would have been willing to sell each additional unit. In perfect
competition, the supply curve shows the amount that firms are willing to
supply at each market price. In long-run competitive equilibrium the
supply curve is the horizontal sum of marginal cost curves of all firms
above their point of minimum average cost. Chapter 5, Section 3.4.
 Competition and welfare

Consumers’ and producers’ surplus in a perfectly competitive market

(Textbook, Fig.7.1). For producers, price P e is equal to the marginal cost
of producing the last unit of output at Q e. Area C = Total cost of
prodn.
Price, cost Producer surplus = area B. S

A
Pe

B D

Qe Quantity lnn
The whole area under the demand curve up to Q e (A+B+C) measures full benefit to
consumers, in money, for consumption of Q e. Consumers pay P e, equilibrium price.
 Total surplus and competition policy, Textbook
p.309.

Producer surplus = area B = industry profit in long-run equilibrium.

Consumer surplus = area A
Total economic benefit = consumer surplus + producer surplus = A + B.
Often weights are given to the importance of benefits given to different
groups.
Heavier weight may be given to consumer benefit (as for the European
Commission, the Ryanair case : Textbook p.310.
Welfare, W = CS + ƛ PS, where, 0 < ƛ < 1. Textbook p.310.

Although competition policy tries to follow principles that should be applicable

in a variety of market situations, in practice competition policy cares about two
interrelated factors: (1) the exercise of market power to raise prices, (2)
manipulation of market structure to undermine incentive to get better.
 Competition policy, market power and market
structure
A firm’s market power is its ability to raise its price above marginal cost. If price is
set above marginal cost, a welfare loss results.
Market structure covers the character of the market including the number and
concentration of sellers, and the distribution of their market shares, and ease of
entry.
Competition authorities find it easier to measure market shares than to measure
marginal costs.
Loss of consumers’ surplus resulting from market power (Textbook, Figure 7.2)

Price Marginal cost

Cost
Pm A
B C
P c = MC c
D E
MC m = MR m
F
Demand

Output
Qm Qc
 Fig. 7.2 Loss of total surplus from market power
Textbook, p.311-312.
This exercise compares the case of supply by a monopolist Q m (single seller) with the
case of supply by a competitive industry Q c.
In a perfectly competitive industry, long-run equilibrium market price P c is the price at
which demand = supply, and where P c = MC c.
At P c, consumers buy Q c of the good.
Consumer surplus under perfect competition is A + B + C.
Producer surplus is D + E + F.
Total surplus = A + B + C + D + E + F.
Under monopoly, the firm has power to set higher price P m, while consumers purchase
Q m. Marginal cost is Q m = Marginal revenue m.
At P m, consumer surplus has shrunk to area A, while
producer surplus = B + D + F.
Total surplus = A + B + D + F.
Setting price at P m, above MC m, reduces total surplus by C + E.
This reduction is deadweight welfare loss: the loss to one group that does not accrue to
another. Textbook p.312.

.
 Market power and market structure
Market power is the ratio of price-cost margin ( the difference between market
price and firms’ marginal cost ) to market price. This is the Lerner Index
(Textbook, pp.312-314).
Textbook p.313. Market structure refers to characteristics of a market that
influence the nature of competition within the market, such as the number and
concentration of sellers. The distribution of market shares tells us whether we
are looking at an industry with a number of firms of similar size , or a market
where a few produce most of the total output.
Market structure affects firms’ incentives to use technology efficiently, and
affects their incentives to invest in better technologies (dynamic efficiency.
Textbook, p.314). Competition authorities may defend innovation, as in the
proposed Dow – Dupont merger, where the producers planned to cut back
research after the merger.
A firm in perfect competition is a price taker. Its demand curve is horizontal. It
has no market power.
A monopolist can either set market price or decide quantity sold. Its
downwards sloping demand curve means P m > MC m, because the marginal
revenue gained from an additional unit sold is less than the price at which it is
sold. To sell an extra unit, then price must be lowered, say to P1 < P0.
Textbook p.317. Monopolist’s marginal revenue MR = P1 – Q0(P0 – P1)
 Cournot model connects number of firms and
market power Textbook pp.317-319)
The Cournot model assumes the market is served by N independent firms, all
of the same size. All produce a homogeneous product.. For industry output
Q0, each firm i produces qi = Q0/N, and each has a market share si = 1/N
Suppose the firms compete by output-setting rather than by price.
Suppose each firm is large enough relative to the market for their individual
output decision to influence market price. Textbook p.317.
(1) The Cournot firm gets additional revenue P1 from selling the extra unit,
but (2) must in consequence sell all of its output at the lower price P1.
MRi = P1 - (Q0/N) (P0 – P1), (Textbook p.318)
The second term on the right-hand side is smaller for Firm i than for the
monopolist. Firm i has less to lose from the price drop required to sell an
extra unit. A firm that is not alone in the market bears only part of industry
loss. Each competitor is too aggressive for the interests of the industry.
So we predict, in aggregate, firms in this industry will find it profitable to
expand output above the profit maximising output of a monopolist.
The larger the number of firms, N, the smaller the loss that each firm suffers.
Cournot provides a negative relation between N and market power (PCM).
Competition policy, concentration & market power
(Textbook pp.319-323)
The loss to a large firm (relative to the market) from a price decrease is large. So
competition authorities pay attention to the number of firms and to the distribution of
market share. The n firms concentration ratio is the simplest index of concentration
and market power: it is the percentage of total industry output accounted for by the
largest n firms in the industry. Textbook, p.319.

The Herfindahl – Hirschman concentration index, HHI, Textbook p.320, has advantages.

HHI = ∑ (Si)squared

i=1
HHI measures the sum of the squares of market share of each firm in the industry.
Its advantages are: (1) it covers the market shares of all firms in the industry, rather
than just a few, (2) The HHI is mathematically related with the Lerner Index by means
of the absolute value of the price elasticity of demand. (Textbook, p.321).
 Entry, market power, SCP model, and efficiency.
Textbook pp.324-326.
Contestable markets are where firms can freely enter and exit the market without
incurring losses when they dispose of their capital. There are no sunk costs. Sunk
costs are a firm’s fixed and irrecoverable costs. They are specific and have no resale
value. Otherwise the incumbents’ market power would be limited by hit and run
raiders.
Competition authorities consider barriers to entry or barriers to expansion of existing
firms in the market. Textbook p.325. Cases are limited availability of inputs,
institutional barriers e.g. planning permission problems, established brands, strategic
barriers. Google used the power of its search engine in its comparison shopping
services to demote rivals’ so on average Google’s rivals lost up to 90% of their traffic.
The Structure, Conduct, Performance Model, SCP, states that Market Structure
(concentration, entry/mobility problems, econs of scale, product differentiation)
influences the behaviour (Conduct) of firms (price setting, investment and innovation,
advertising, collusion), which determines the Performance of the industry ( welfare
impact, firms technical or dynamic efficiency). On the contrary, the ‘efficiency
hypothesis’ (Demsetz), p.326, counters with the argument, efficient firms have above
average profits.
Market structure affects efficiency of the industry through incentives to avoid X-
inefficiency (Textbook pp.217-8, 326) . Competition gives an incentive to limit X-
inefficiency.
 Competition and strategic behaviour
Module textbook, Chapter 7, Section 4.

Oligopoly is a market containing a small number of firms whose

behaviour is influenced by recognised mutual interdependence.
A profitable dominant firm risks attracting new profit-seeking entrants.
Such a firm may therefore decide to over-produce, driving down market
price to below a limit price (maximum price an incumbent firm can set
that limits new entry to zero).
Game theory allows us to explore the strategic reasoning of both the
incumbent and the firm considering market entry. E.g. Fig.7.5, p.329 (A
three-stage entry game in extensive form). Also you may wish to refer to
Fig.6.15, p.279, which structures such a multi-stage game into three
subgames. Backward induction is a method of solving games that looks
ahead to the final payoffs and works backwards to the beginning of the
game. Module textbook, p.275.
Fig.7.6, Textbook p.332, models a game where the incumbent achieves
entry deterrence by investing in sunk costs.
 A taxonomy of anti-competitive behaviour
Module textbook, Chapter 7, Section 5.
Business strategies to give firms power over competitors and leading to
high profits are likely to be at odds with competition law.
However some may not be welfare-reducing. E.g. Product differentiation
may better meet consumers diverse preferences.

Strategies that are likely to fall foul of competition law are :

(1) collusion, p.335 (agreements to restrict competition, punishment,
uncertainties), (2) horizontal mergers, p.336 (perhaps supported by
arguments for efficiences), (3) abuse of dominant position, p.338
(hampering ability to compete or removing some competitors), (4)
predatory pricing, p.339 involving setting prices below cost for long
enough to induce exit of some competitors, followed by price rises, an
action enabled by greater financial resources. A profitable track record
attracts finance.
 Chapt. 5.2, 5.3, 5.4, 5.5 Technical change: from static to dynamic
efficiency

Productive Efficiency is a key neoclassical notion. 3 types of static efficiency (assume

technology is fixed) that apply to the supply side of markets are:
(1) Firms should produce with technical efficiency. P.201. (2) Firms should minimise costs for
each level of output. P.205. (3) Firms should produce at an efficient scale: at minimum LRAC.
P.208-210, Figure 5.13.
Technical change can be incorporated into the neoclassical model as shifts in the production
function (Fig.5.16 and Fig.5.17 Cost minimisation before and after technical change. P.216)
Dynamic efficiency is the increase in technical efficiency that occurs over time with the
introduction of technological innovation (p.216). The firm’s expansion path and related
cost functions will change.
Neoclassical theory models technical change as exogenous, and therefore ignores the
sources of innovation. Nelson and Winter (1982) proposed an evolutionary framework of
processes by which variables in a system change over time (replacing models of market
equilibrium), where continuous industrial change is driven by innovation (novelty: new
products, new ways). Innovation is key to modern capitalist economies. Evolutionary
concepts: bounded rationality; satisfycing; routines; tacit knowledge; path-dependence:
Textbook pp.224-226. Like Hayek, Schumpeter saw markets as discovery procedures
(endogenous), ‘creative ..’.
Temporary profits provide funds, reward innovation. Models should include:
•persistent heterogeneity & disequilibrium • the importance of ‘selection’ of firms • key role
of innovation & entrepreneurship • trial, error and failure.
 Dynamic efficiency, competition and policy
A technique of production is defined by a fixed proportional relation between inputs: e.g.
K/L. Textbook, p.194.
A firm is technically efficient if it is using the minimum quantity of inputs needed to
produce a given output for each technique of production. P.201,
Dynamic efficiency is the increase in technical efficiency that occurs over time with the
introduction of technological innovation (in process and products). P.216.
Market structure affects dynamic efficiency. Competition authorities need to be careful
that in scrutinising market power, they do not remove incentives for R&D and technical
improvement.
Is R&D (improving technologies) higher in highly concentrated markets than in very
competitive ones? The answer depends upon firms’ ability and incentives to invest.
With regard to ability to innovate in concentrated markets, such markets may be more
conducive to innovation because information is imperfect (uncertainty and asymmetric
information about rewards). Schumpeter.
With regard to incentives to invest in innovation, there is a mixed picture. Desire to erect
protective barriers to entry may drive innovation by powerful firms. Network effects may
be important. But the incentives confronting a new entrant who has ‘nothing to lose’
may be large compared with firms whose existing technology is threatened by
innovation.
Competition policy is thus a complex balancing act, sustaining incentives for efficiency
while undermining market dominance that resists welfare-enhancing competition
(Regibeau, Module textbook, p.344).
 Part C: Subtle shadings in approaches to regulation of pharmaceutical
companies, in recognition that higher profits from these firms can fund
innovation and ESG.
Note that there is a case to be made for both competitive markets and innovating firms.
Therefore regulators such as the CMA – UK’s primary competition and consumer authority –
may allow innovating firms to increase market share, which may be against the ideals of
competitive market theory reliant on the concept of deadweight welfare loss (Module
textbook, Chapter 7, Section 2.2).
Competition authorities may focus more on the importance of consumer welfare than on
firms’ profits. (Chapter 7, Section 2.1) We may illustrate the theory of collusion, mergers and
dominance with the CMA articles, together with materials on anti-competitive behaviour.
Note the tension between regulator’s focus on profits, and risk of lessened incentives for
R&D, and technical improvement.
Evolutionary theory (Chapter 5, Section 5.3) may provide a case for incumbents who increase
both concentration of markets at the expense of new innovating entrants and productivity.
We may use the CMA articles to highlight these tensions. Incumbents can increase economic
welfare.
Social welfare is a broader concept than economic welfare . It includes indirect and long-term
effects, such as Oxford/AZN’s concern towards equitable distribution and affordability of
vaccines, as well as direct effects, as emphasised by Pfizer/BioNTech) and Moderna.
Oxford/AZN appear to be led by clear long-term reputational and ESG related benefits, as
they work to coordinate vaccine delivery to low-income nations.