Where
t
denotes inflation,
t
y denotes the log deviations of real GDP from its long run
trend, and
t
denotes the error term. However, it is more advanced version of the curve.
In this case, lets begin with the original Phillips curve, published by A. W. Phillips in
1958. This relationship was basically an empirical finding looking for the theoretical
background. The Phillips curve has been an empirical finding in search of a theory, like
Pirandelo characters in search of an author (Tobin, 1972). Initially, the Phillips curve was
viewed as a relationship between the rate of change in money wages and the level of
unemployment. Using the data from Lithuanian department of statistics, the following figure
shows how well data from the sample period (1998-2009) fits the original Phillips curve
relationship in case of Lithuania.
Earnings and unemployment rate in 1998-2009
-10.00%
-5.00%
0.00%
5.00%
10.00%
15.00%
20.00%
25.00%
0 5 10 15 20
Level of unemployment (%)
I
n
c
r
e
a
s
e
i
n
m
o
n
e
y
w
a
g
e
s
Figure 3.1. Phillips curve (1) in Lithuania during the period 1998 to 2009
Despite many empirical failures of this type of Phillips curve, it looks reasonably similar
to the original relationship found by Phillips in 1959. Only two red dots (which
correspond to two pairs of wage inflation and unemployment) are a little bit distant from
the trend line (green line). Phillips (1958) used a mathematical expression to generalize
his empirical findings. The latter had a form ) (U f w =
-
, where w is logarithm of money
wages (a dot over a variable indicates rate of change of the variable) and U is the level of
unemployment. The logic is trivial: the higher shortage of labour we have, the quicker we
need to increase its price. The empirical relationship as we can see in figure 3.1 was
given a theoretical interpretation by Lipsley (1960). In his model in unemployment was a
(negative) proxy for excess demand for labour and wage change responded to excess
22
demand for labour (Arestis, Sawyer, 2007). Despite further developments made by
Friedman (1968) and other researchers, Lipsleys formulation of the Phillips curve
became old-fashioned, although it offered the initial theoretical justification of the model.
It is often concluded that there is no firm basis to state that there is a negative relationship
between the rate of change of wages and the level of unemployment. This is a fair
statement to finish up with this version of the Phillips curve and move to the other,
slightly different one.
Lets have a look at different version of the Phillips curve a relationship between price
inflation and unemployment. For this purpose, a figure below is presented.
Inflation and unemployment in 1998-2009
-2.00%
0.00%
2.00%
4.00%
6.00%
8.00%
10.00%
0 5 10 15 20
Level of uneployment (%)
A
n
n
u
a
l
i
n
f
l
a
t
i
o
n
r
a
t
e
a
c
c
o
r
d
i
n
g
t
o
t
h
e
c
o
n
s
u
m
e
r
p
r
i
c
e
i
n
d
e
x
Figure 3.2. Phillips curve (2) in Lithuania during the period 1998 to 2009
Slightly worse, but still a visible fit of the Phillips curve can be seen in figure 3.2. No
signs of stagflation (combination of high inflation and high unemployment) like were in
1970s. Samuelson and Solow (1960) contributed a lot in translating the original Phillips
curve to this one, which relates the price inflation with unemployment. Equation of this
macroeconomic model usually is expressed using the following form
(3.2) ) ( y g p p
e
+ =
-
-
where y is a measure of output gap.
Using wage inflation and price inflation as closely related variables basically implies that
inflation is an unemployment related issue instead of being related to the product market.
However, Arestis and Sawyer (2007) argue that there is the question of the compatibility
of mark-up pricing with the wage inflation Phillips curve. Their argument is Walras law
which states that excess demands over the markets equal to zero. Appealing to this law, if
we had an excess demand for labour, then we would have an excess supply of output in
the product market. Arestis and Sawer (2007) continue: Positive excess demand for
labour with real wages below equilibrium would lead to wages rising faster than prices,
and that would be complemented by negative excess demand for output with price/wage
ratio above equilibrium and prices tending to fall relative to wages. In other words,
according to Walrass law, wage inflation and price inflation should not be closely
relates. Obviously, this result holds if and only if the Walrass law holds. However, it is
23
claimed that Walras law is often violated by Keynesian macroeconomic models
(Sargent, 1979). This is also one of concerns about the plausibility of Phillips curve if it
violates Walras law.
In addition, Arestis and Sawyer (2007) underlined a few other shortcomings of the
Phillips curve as a relationship between price inflation and unemployment, some
developments, but they are not under scope of this dissertation. Summing up, they
conclude that the excess demand approach relating to the Phillips curve is problematic
and does not provide a secure basis for it. Moreover, this approach designed for the
presence of perfect competition, when the supply and demand curves are defined. When
we have imperfect competition, firms are price makers and by setting their prices these
firms eliminate any excess demand. Arestis ans Sawer (2007) says that under imperfect
competition, the level of price (relative to cost) is related (though not monotonically) with
output, and hence it appears that the rate of change of price will be related with the rate
of change of output under imperfect competition. But new Keynesian theory has a
different view about this. Leaving two above mentioned approaches behind, we now
move to the new Keynesian approach.
3.2 The new Phillips Curve and its Shortcomings
Lets assume that we are going to determine a model in an environment of
monopolistically completive firms that face some type of constraints on price adjustment.
One of the constraints follows the scenario suggested by Taylor (1980) in his work about
the staggered contracts. The constraint is that the price adjustment rule is time dependent.
In the spirit of Taylors (1980) model, assume that all firms change their prices for N
periods of time. A constant fraction
N
1
of all firms changes their prices in any given time
period. This example differs from the Taylors model in the way that the pricing decision
evolves explicitly as a result of profit maximization problem (of monopolistic
competitor).
Gali and Gerler (2008) remind that such aggregation as defined by Taylor (1980) is
indeed a complicated task to do: It is necessary to keep track of the price histories of
firms. Thankfully, we can use an assumption originally raised by Calvo(1983) and now
commonly used. This assumption simplifies the aggregation problem by a large-scale.
The key elements of this assumption from Calvos (1983) staggered price model are as
follows. In each period some firms change price and others do not. In each period, a
fraction of firms do not change their price. To put it differently, in each given period,
each firm changes its prices with probability 1 and leaves unchanged with
probability . This probability does not depend on the time elapsed since he last price
adjustment. According to Gali and Gertler (2000), the expected time when the price
remains fixed equals to
1
1
) 1 (
0
1
k
k
k . In such way, the parameter provides a
measure of price rigidity. For example, if 75 . 0 = (in a quarterly model), then firms keep
their prices fixed on average for a year. By using this approach we greatly simplify the
24
aggregation problem, because the adjustment probabilities do not depend on firms price
history.
Assume that all firms are identical ex ante. The only differences are that each firm
produces a differentiated product and firms have different price histories. In addition,
assume that each firma faces a downward sloping constant price elasticity curve (Gali
and Gertler, 2000). Then, applying the law of large numbers and log-linearizing the price
index around a zero-inflation steady state, it is possible to construct the aggregate price
level using a weighted average of lagged price level
t
p and the optimal reset price (or the
optimal changing price)
*
t
p , as follows
(3.3)
*
1
) 1 (
t t t
p p p + =
her p denotes the (log) price level and
*
p denotes (also log) the newly set optimal price.
Considering that all firms are identical (except the differentiate products they produce), at
time t a fraction 1 of firms should set their prices at new level
*
t
p . Appealing to the
law of large numbers, the index of prices of those companies which do not change prices
during the period is simply equal to the lagged price level (Gali and Gertler, 2000). It is
not difficult to show that profit-maximizing firm will set
*
t
p according to the following
log-linear rule:
(3.4) { }
n
k t t
k
k
t
mc E p
+
=
0
*
) ( ) 1 (
Where is a subjective discount factor and
n
k t
mc
+
is the logarithm of nominal marginal
costs in period t+k of the firm that last time set its price in period t. This approach defines
the optimal reset price that is set by profit-maximizing firm, taking into account time
dependent pricing rules suggested by Calvo. In the limiting case ( 0 = ) we have perfect
price flexibility and the firm changes price proportionately to movements in the current
marginal cost (Gali and Gertler, 2000). When the degree of price rigidity increases (this
degree is measured by ), then the expect time when the price is likely to remain fixed
also increases. As a result of this, the firm puts more weight on expected future marginal
costs when choosing the current price (Gali et al, 2001).
Assuming that inflation at period t can be expressed as
1
=
t t t
p p , the percent
deviation of the firms real marginal cost from its steady state value denoting as
t
mc , and
combining equations (3.3) and (3.4) we can get the following expression:
(3.5) { }
1 +
+ =
t t t t
E mc
where
) 1 )( 1 (
. Note that the latter depends on the frequency of price
adjustment and subjective discount factor . Rewriting equation (3.5) for next period
we get
25
{ }
2 1 1 + + +
+ =
t t t t
E mc
(3.5a) { }
3 2 2 + + +
+ =
t t t t
E mc
{ }
4 3 3 + + +
+ =
t t t t
E mc etc.
Substituting equation (3.5a) into equation (3.5), we obtain that inflation today
corresponds to the expected discounted marginal cost flow (Melihovs, Zasova, 2007):
(3.6) { }
=
+
=
0 k
k t t
k
t
mc E
Consequently, this means that firms make pricing decisions based on expectations of the
future behavior of marginal costs. This conclusion comes intuitively knowing that firms
are forward looking, they prices as a markup over a discounted stream of expected future
nominal marginal costs, and must lock into price for multiple periods (Gali and Gertler,
1999).
According to Gali and Gertler (1999), we can assume that marginal costs are proportional
to the output gap (a log-linear relationship exists). In addition, denoting
*
t t t
y y x = as
output gap (where
t
y and
*
t
y denote the natural logarithms of actual output and potential
output respectively), we can express the marginal cost as a function of
t
x :
(3.7)
t t
kx mc =
Where k is the output elasticity of marginal cost
Combining the relationship between marginal cost and the output gap (equation (3.7))
with equation (3.5), we obtain:
(3.8) { }
1 +
+ =
t t t t
E kx
The latter equation shows that the new Phillips curve implies that current inflation
depends on output gap and inflation expectations, so it is similar to the traditional Phillips
curve. However, unlike the in the standard Phillips curve, ) (
1 t t
E
is replaced
by ) (
1 + t t
E . That is, actual inflation is affected by currently expected inflation of next
periods rather than lagged inflation of previous periods (Melihovs, Zasova, 2007).
Despite the new approach, this type of Phillips curve, also known as new Keynesian
Phillips curve, faces some shortcomings in theoretical and empirical considerations. Gali
and Gertler (1999) emphasizes that the conventional output gap measures may not be
particularly reliable due to unobservable rate of natural output
*
t
y . They argue that despite
various proxies for the potential output
*
t
y , these measures contain significant
measurement error. Another issue mentioned by Gali and Gertler regards the conditions
26
under which the output gap corresponds to marginal cost. They are concerned if these
conditions are satisfied. And if not, structural estimates of the Phillips curve based on the
output gap should be treated with caution. Another implication of the NKPC is also
underlined by Gali and Gertler (1999). According to them, equation (3.5) implies a
negative relationship between the lagged output gap and current change in inflation rate;
however, their study with the U.S. data has shown that the inflation rate depends
positively on the lagged output gap rather than negatively: The estimated equation,
unfortunately, resembles the old curve rather than the new! (Gali and Gertler, 1999).
Arestis and Sawer (2007) add some more criticism on the NKPC. They argue that
although derivation of equation (3.4) is based on the assumption of increasing marginal
costs, results from various surveys show that most of the respondents confirm the
opposite declining marginal costs when producing additional units of product. In
addition to many more arguments (see paper for details), Arestis and Sawer (2007)
conclude that the NKPC based on the output gap lacks theoretical justification.
3.3 The new Keynesian Phillips Curve by Gali and Gertler
After all the criticism presented by different authors regarding the NKPC using the output
gap measures, Gali and Gertler (1999) offered a new approach. The idea of their work
was to express inflation in terms of an observable measure of aggregate marginal cost. In
other words, they estimate equation (3.5) instead of equation (3.8). Gali and Gertler use
restrictions from theory to derive a measure based on observable marginal cost.
Lets begin with derivation of real marginal cost measure. One of the simplest measures
of marginal cost is one based on the assumption of a Cobb-Douglas technology. Denoting
technology with
t
A , capital with
t
K , and labour with
t
N , output
t
Y is given by
(3.9)
n k
t t t t
N K A Y
=
The ratio of wage rate to marginal product of labour then corresponds to real marginal
cost and is expressed as
t t t
t
t
N Y P
W
MC
c c
=
/
1
. Therefore, given equation (3.9) we have:
(3.10)
n
t
t
S
MC
=
Where
t t
t t
t
Y P
N W
S is the labor income share. Using lower case letters for denoting the
percent deviations from the steady state we have:
(3.11)
t t
s mc =
Combining equations (3.11) and (3.5) we get the following expression for inflation:
27
(3.12) { }
1 +
+ =
t t t t
E s
Where the coefficient is given by
(3.13)
) 1 )( 1 (
Taking in account that under rational expectations there is no correlation between the
error in forecast of
1 + t
and information dated t and earlier, we can derive from equation
(3.11) that
(3.14) { } 0 ) (
1
=
+ t t t t t
z s E
Where
t
z denotes a vector of variables dated t and earlier. Equation (3.13) refers to the
orthogonality condition and forms the basis for estimating the model via Generalized
Method of Moments (Gali and Gertler, 1999).
4. What will be done next
In fact, I am having some difficulties in the process of estimation, but hopefully I will
manage it. Obviously, thats why I am sending not a full paper.
No introduction yet, but it will go lastly. I will also add what has been done by other
researchers regarding the Phillips curve in Lithuania. I did not write about the hybrid
model, because it might be above my abilities and the main idea of this job is to write a
paper which mainly gives basic ideas of what is the Phillips curve and how it became so
different in comparison with the original one. In fact, I am doing not very advanced, but
rather important job to me and maybe to someone else who will be interested in my
native country. That is why I wrote so much about the economic development and all
main processes that were in those 20 years of independence. By going this way I
managed to distance from simply copying other authors and ideas and giving some
simple analysis by myself. Being a student of investment studies, it was dangerous to
choose such topic, but now I am happy anyway, despite the mark I am going to get,
because I learned interesting about macroeconomics, got familiarized with most
important economic figures, etc.
Any comments of yours will be highly appreciated.
References.
[1] A. W. Phillips, The Relation between Unemployment and the Rate of Change of
Money Wage Rates in the United Kingdom, 1861-1957. Economica, New Series,
Vol. 25, No. 100 (Nov., 1958), pp. 283-299.
[2] Kevin D. Hoover, Phillips Curve. Available at
http://www.econlib.org/library/Enc/PhillipsCurve.html. Accessed at 15/07/2010.
28
[3] Phelps, Edmund S. Phillips Curves, Expectations of Inflation and Optimal
Employment over Time. Economica, n.s., 34, no. 3 (1967): 254281.
[4] Friedman, Milton, The Role of Monetary Policy. American Economic Review 58,
no. 1 (1968): 117.
[5] Lucas, Robert "Econometric Policy Evaluation: A Critique", in Brunner, K.; Meltzer,
A., The Phillips Curve and Labor Markets, Carnegie-Rochester Conference Series on
Public Policy, 1, New York: American Elsevier, (1976), pp. 1946
[6] Robert J. Gordon, What Is New-Keynesian Economics?, Journal of Economic Literature,
Vol. 28, No. 3 (Sep., 1990), pp. 1115-1171
[7] Saleh M. Nsouli, A Decade of Transition. An Overview of the Achievements and
Challenges, Fianance and Development, a quarterly magazine of International
Monetary Fund, June 1999, Volume 36, Number 2.
[8] Thomas Grennes, The Lithuanian economy in transition, Lituanus, Lithuanian
Quarterly Journal of Arts and Sciences, Volume 40, No.2, 1994.
[9] World Bank. Lithuania: The Transition to a Market Economy. Washington, D.C.:
1993.
[10] Department of Statistics to the Government of the Republic of Lithuania (Statistics
Lithuania), available at www.stat.gov.lt
[11] Melihovs A. and Zasova A., Estimation of the Phillips curve for Latvia, Bank of
Latvia, 2007.
[12] Ober-Haus real estate advisors. All information used in the dissertation is available
at http://www.ober-haus.lt/news/price-index. Last time accessed at 25/07/2010.
[13] IEA: Oil supply crunch and mega-recession by 2013, available at
http://www.whatmattersweblog.com/2009/03/01/iea-oil-supply-crunch-and-mega-
recession-by-2013/. Last accessed at 28/07/2010.
[14] Tobin, J. Inflation and Unemployment, American Economic Review, (1972),
62(1), 1-26
[15] Samuelson, P. and Solow, R.,Analytical aspects of anti-inflation policy, American
Economic Review, (1960), vol. 50(2), pp. 177-94
[16] Sargent, T.J., Macroeconomic Theory, New York:Academic Press, 1979.
[17] Thomas I. Palley, Walras' Law and Keynesian Macroeconomics, 1997.
[18] Gali J. and Gertler M., Inflation Dynamics: A Structural Econometric Analysis,
2000.
[19] Gali J., Gertler M., and Lopez-Salido D., European Inflation Dynamics, 2001.