1 Seminar on Inflation Persistence in Latin America learning from by tyndale


									  Seminar on Inflation Persistence in Latin America: learning from research in the
                                     Euro Area
                                August 23rd and 24th
                             Central Bank of Argentina

Brief summary of presentations and discussion

With the aim of stimulating interactions among central bank economists in the region
developing research on the issue of inflation dynamics and learning from research
developed in the Euro Area, the Central Bank of Argentina invited a few participants of the
Inflation Persistence Network (IPN) to present and discuss the main results produced under
this initiative. The IPN is a joint research effort of the European System of Central Banks
(ESCB) to investigate the issues of inflation persistence and price formation in the Euro
Area and their policy implications. Economists of different Central Banks in Latin
America working on these issues were also invited to present their contributions at the

A rich debate was motivated by presentations of both IPN members and the local
presenters. In what follows we briefly summarize the main points discussed during the

Andrew Levin opened the first day sessions presenting a brief summary of the main
findings of the IPN regarding inflation persistence in the Euro Area and other industrialized
countries and its relationship with regime shifts. He shed light on the difference between
persistence as a statistical concept and its meaning in the context of monetary economics.
While the former mainly refers to the speed of reversion towards a sample mean, the
economic definition of persistence should consider the presence of economic policy aiming
at stabilizing inflation at a desired level, which in turn reflects some perception about what
the long run equilibrium level of inflation is. Under this approach shocks to inflation are
not expected to have a permanent effect.

The literature on inflation dynamics has relied on the statistical approach to measure
persistence for many years and the finding of inflation being nearly a random walk was not
surprising. Moreover, high persistence was considered a distinctive characteristic of
inflation dynamics. One of the main findings of the research developed by the IPN for
Europe and other developed countries regarding persistence at the aggregate and sectoral
level, is that once shifts in the monetary regime are taken into account by considering the
possibility of changes in the long-run equilibrium (mean) level of inflation, estimated
persistence decreases significantly. The successful disinflation experiences of the US and
other developed economies in the 90’s and afterwards in which the Fed was able to reduce
the medium and long term value of inflation is also clear in this respect.

The findings presented for some Latin America countries regarding inflation persistence at
the aggregate level are mostly in line with those of the IPN: (i) there is evidence about
coincidence of regime shifts with breaks in mean inflation (ii) there is in general a
significant reduction in estimated persistence once changes in mean inflation are considered

and (iii) the generalized inflation slowdown in the region since the beginning of the 90’s
has been accompanied in many of the cases by a reduction in aggregate persistence (iv) the
effects of the adoption of inflation targeting by some of the countries in the region on
inflation dynamics are rather unclear.

Regime shifts have been however of major magnitude in the region, compared to those
experienced by industrialized countries. In particular, some countries experienced very high
inflation during the 80’s and in some cases hyperinflation episodes by the end of this
decade. During those years and for those countries, monetary policy was exogenously
driven because of fiscal dominance and it is not straight forward that it was playing an
effective role in stabilizing inflation towards a certain long-run equilibrium level. In recent
years, countries in the region made an important effort towards more sound fiscal policies
and monetary policy seems to have recovered the role of providing a nominal anchor to the

In this respect, the paper by Carlos Capistrán and Manuel Ramos Francia, from the Banco
de México, studies the inflation process in the ten largest Latin American countries over the
period 1980-2006. The authors find that although persistence remains relatively high for
some countries relative to industrialized countries, the most relevant feature of the inflation
process in the region is the decline in mean inflation over recent years. For at least five of
the ten countries persistence is currently lower compared to historical standards. While the
removal of fiscal dominance seems to have played a role in reducing mean inflation and to
some extent persistence, the evidence about the changes brought about by the adoption of
inflation targeting is less clear.

Quite similar results emerged from one of the papers by Laura D’Amato, Lorena Garegnani
and Juan Sotes Paladino of the Banco Central de la República Argentina, which focuses on
inflation persistence at the aggregate level. Univariate time-series analysis is used to study
inflation persistence over the period 1980-2007. Significant changes in the mean and
persistence of inflation are identified, which are clearly related to regime shifts: the
hyperinflation episode in August 1989 and the abandonment of the Convertibility regime in
January 2002. When a time varying mean is subtracted to inflation, the estimated
persistence decreases significantly. Inflation was highly persistent during the high inflation
period, but it strongly declined when inflation lowered after the adoption of Convertibility
in 1991. Then inflation persistence increased slightly but nevertheless significantly after the
adoption of a managed float in 2002. The results of the paper are in line with the findings of
a clear link between changes in mean inflation and regime shifts. It also gives evidence of
significant changes in persistence related to those breaks.

The variance of inflation could also be influenced by regime shifts and this could be
particularly important for developing economies, that have experienced high
macroeconomic volatility. The transparency of the Central Bank in transmitting its policy
objectives to the private sector is crucial in this respect. Once private agents incorporate
this objective to their expectations, inflations tend to converge to the Central Bank target
and its volatility lowers. The empirical evidence suggests that, at least for developed
countries, explicit targets have played a key role in successfully anchoring inflation
expectations and reducing inflation persistence.

An issue of discussion was whether it could be reasonable to model inflation as a random
walk for periods in which the national Central Banks was not able to stabilize inflation.
There was quite agreement on the fact that even for those cases it should no be reasonable
to model shocks to inflation as having a permanent effect on it.

Something noticeable about the modeling of inflation dynamics for Latin American
economies is that it departs in some way from that used for industrialized countries,
because of the need to deal with huge structural breaks and high heteroskedasticity. In
particular, the two papers presented by Alberto Humala, from the Banco de la Reserva del
Perú, model inflation as a process subject to regime switches.

One of these papers by Paul Castillo, Alberto Humala and Vicente Tuesta studies in detail
the link between inflation and inflation uncertainty in a context of monetary policy regime
shifts for the Peruvian economy. The inflation time series is decomposed into its permanent
and transitory components in order to establish the link between inflation and inflation
uncertainty (both at long and short run). Regime switching behavior in the variance of
shocks to the permanent and transitory components of inflation is allowed (into a Markov
switching heteroskedasticity model of inflation) to disentangle influence from policy
changes. They conclude from their results that keeping trend inflation under control and
dragging inflation expectations down reinforces credibility in a central bank’s inflation-
intolerant policy. This, in turn, reduces the volatility of both permanent and transitory
shocks by lowering inflation expectations and reinforcing the credibility in central bank’s
actions. Reduction in persistence and thus in stabilization costs, might be due to fall in
both, long-run and short-run uncertainty.

The other paper, by Alberto Humala applies the same methodology to study whether or not
the volatility of permanent and transitory shocks to inflation in the ten largest South
American countries has decreased. He is able to identify two different patterns among the
ten countries. For those that had experienced hyperinflation episodes as Argentina, Bolivia,
Brazil, Chile and Peru, three regimes can be inferred: one of low and stable inflation, other
of high and volatile inflation and a hyperinflation regime. The other group corresponds to
the non-hyperinflation countries: Colombia, Ecuador, Paraguay, Uruguay and Venezuela.
For this last group, he identifies two regimes: one of low and stable inflation and another
under which inflation is high and volatile. While both groups of countries experienced
regime switches in inflation, countries in the first group seem to have reached price stability
earlier and for longer periods. On the contrary, countries that have not experienced
hyperinflation episodes appear to be more reluctant to bring inflation down permanently.
The observed slowdown of short-term shocks to inflation could be an effect of the decrease
in inflation expectations, while long-term shocks volatility might have decreased due to
more inflation-intolerant monetary policies.

Going one step down in the level of aggregation, there is evidence that sectoral inflation
does not depart fundamentally from aggregate dynamics. In fact, the paper by Tomás
Castagnino, Laura D’Amato and Juan Sotes Paladino, from the Banco Central de la
República Argentina, finds evidence of breaks in sectoral inflation dynamics which are, as
overall CPI inflation, associated to the regime shift in January 2002. The abandonment of

the currency board and posterior adoption of a managed float was found to be linked to a
rather widespread raise in mean, variance and persistence of inflation across sectors. In
addition, despite of the fact that inflation persistence seems to be heterogeneous and much
lower than in the aggregate, only one common factor helps to explain a lot of the long run
dynamics of individual series. This result is mainly due to the aggregation process through
which the idiosyncratic, high frequency components are lost, while common, low frequency
ones remain. What is more, this shared factor tracks quite well aggregate dynamics, what
results in a potentially good, structurally built, non ad-hoc measure of core inflation.

Regarding price formation, Philip Vermeulen presentation focused on the main findings of
the IPN research based on micro PPI data. Some important regularities appeared from
research across different countries and studies: (i) price changes are rather infrequent and
relatively large compared to average inflation, (ii) price setting behavior is heterogeneous
across different sectors, but sectoral prices behave quite homogenously across countries,
and (iv) prices do not seem to exhibit downward rigidity, contrary to conventional wisdom.
Among the main determinants of price stickiness, higher labor shares are associated to a
lower frequency of price adjustment, while a high share of energy increases the frequency
of price adjustment. High-inflation environments are positively correlated to faster price
adjustments, while more competitive sectors are less sticky due to larger opportunity costs.
The frequency of price adjustment is also influenced by seasonality, which may be time-
dependent (winter and summer sales) or state-dependent (wages change when unions get
together). Finally, there is also evidence that CPI prices are stickier than PPI, a result that is
robust to basket, sub-basket and individual product comparison between CPI and PPI.

The paper Juan Pablo Medina, David Rappaport and Claudio Soto, from the Banco Central
de Chile, tried to answer quite similar questions looking at CPI micro-level data in Chile.
The frequency of price adjustment, the size of price adjustment, the price dispersion, and
the synchronization of price movements across different establishments were analyzed.
Their results indicate that prices in Chile are adjusted more often than in some developed
countries of Europe and in the USA. On average, price changes per product take place
every three months. When they investigate the driving factors of price stickiness, they do
not find a clear correlation between price adjustment frequency and the aggregate inflation,
although price adjustment appears to be correlated with inflation when decomposing price
changes in upward and downward price adjustments. They do not find a clear pattern of
interaction between price dispersion and inflation, but they observe some degree of
synchronization in the price adjustment mechanisms across establishments. Finally, they
conclude that their results tend to confirm the view that changes in prices should be
characterized, in Chile, by state-dependent price models rather than time-dependent

Going deeper on the issue of price formation, the presentation by Roberto Sabbatini, from
the Banca d’Italia, focused on the summary results of the IPN coming from survey analysis
conducted to study firms’ pricing behavior across 9 different countries in Europe, which
represent 94% of the Euro Area GDP. In spite of the presence of idiosyncratic features, the
results obtained from this analysis are quite robust. The prevalent market structure seems to
be one of monopolistic competition, in line with the most widely used assumptions to
model pricing behavior in modern macroeconomics and monetary policy analysis. There is

also evidence that when setting and revising prices, firms follow a combination of state and
time-dependent rules and use a wide set of information (past, present and future). Contracts,
coordination failures and lack of competition play a role in adding stickiness to price
formation. Finally, prices seem to respond asymmetrically to shocks: prices increase in
response to cost shocks and decrease when market conditions weaken.

Andrew Levin discussed the implication of the findings on inflation persistence and price
formation for monetary policy modeling focusing on three main points: (i) the empirical
relevance of intrinsic persistence and its implications for monetary policy, (ii) alternative
specifications of real rigidities lead to different policy implications in terms of the output
costs of disinflations (iii) the role played by credibility and transparency of monetary policy
in reducing the costs of disinflations.

Inflation persistence and stickiness in prices are key subjects in conducting monetary
policy. Knowledge about the degree of inflation persistence is important because it
provides the Central Bank with vital information to reach the desired level of inflation.
While a lot of attention has been devoted to explain intrinsic persistence, which is related to
nominal rigidities, less research has been devoted to study the sources of real rigidities.

However, when structural models are used to analyze persistence, a low level of intrinsic
persistence is found in them, assuming partial indexation in wages and prices. By
introducing real rigidities in those models, it can be seen that different sources of real
rigidities, i.e. firm specific factors and quasi-kinked demand, generate very different
steady-state costs and volatility of inflation. In particular, the presence of firm specific
factors produces relative price distortions that have a significant impact in term of output

But the real costs of disinflation are also determined by the particular characteristics of
monetary policy: (i) the horizon for achieving the nominal target, (ii) the operational
strategies given the preferences of the policy maker and the constraints of the economy and
(iii) the transparency and clarity of policies to private agents and (iv) the credibility of

Historical evidence suggests that the credibility and transparency of monetary policy play a
fundamental role in determining the output costs of a disinflation policy. Comparing the
three biggest episodes of disinflation in the United States: the post-Civil War deflation, the
post-World War I deflation, and the Volcker disinflation in the early 80’s, it can be
assessed that the very different macroeconomic effects of these three monetary contraction
episodes are related to the also different design and transparency of the policy strategy
adopted in each case. While the low real costs of the post-Civil war deflation of the 1870s
seems to be related to the highly predictable nature of the price decline, a more predictable
policy of gradual deflation could have helped to avoid the sharp post-World War I
contraction. Regarding the Volker disinflation, a gradualist approach under low credibility
would have made the adjustment more painful by extending the period of disinflation.
Thus, it seems that a gradualist approach seems to have important benefits under a policy
regime with relatively high credibility. On the contrary, when the credibility of the policy
institution is relatively low, a more aggressive change in policy stance makes the policy

shift more evident to private agents. In particular, the analysis of the Volcker disinflation in
the 80’s reveals that when monetary policy lacks credibility, an aggressive policy stance
can play an important role in making a policy shift more apparent to private agents, helping
expectations to adjust rapidly.

Some interesting conclusions for future research agenda in the region came up from the
discussion developed during the two days of the seminar. First, although research on
inflation persistence and price formation in Latin America is still rather preliminary and
disperse, some of the main findings of the IPN at the aggregate level seem to hold for the
region: (i) intrinsic persistence is not so high once breaks in mean inflation are taken into
account when measuring persistence (ii) inflation persistence seems to decrease when mean
inflation lowers (iii) there is a clear link between breaks in mean inflation and regime shifts.
Second, the importance of transparency and credibility in anchoring expectations and
reducing inflation volatility was also emphasized for both industrial and developing
economies. Third, inflation dynamics has some particular features in the region, probably
related to the high magnitude of different kind of shocks to which the region is exposed.
But while high macroeconomic instability was a characteristic feature of the region over
past decades, it seems to have been overcome in the last fifteen years, due to more sound
and robust fiscal and monetary policies. Fourth, imperfect competition and mark-up pricing
seem to be relevant to explain price stickiness at least for Europe, where a deeper
knowledge of the price formation process was achieved through firms survey analysis and
micro level analysis of the CPI and PPI. Research at the micro level is much less developed
in Latin America and seems to be an important part of the research agenda for the region.
Finally, and in connection with this, the initiative of a Latin American Network that could
join Central Bank research economists working on these issues in the region, with the aim
of improving interactions, homogenizing methodologies, and through this, extracting some
regularities for the region was supported by participants. In this regard, institutional
interactions and technical support from the IPN was a possibility that was also perceived as
important to improve the knowledge on inflation dynamics in the region.


To top