PRICE SETTING BEHAVIOUR AND PRODUCT INFLATION IN
LESOTHO
Mamello Nchake
University of Cape Town, South Africa
Abstract
This paper uses a unique data set on monthly product prices at the retail level to analyse
price setting behaviour within Lesotho since 2002. The results reveal substantial heterogeneity in
the frequency, duration and size of price changes across retail outlets, region (rural vs. urban)
and disaggregated product categories. Further, the paper reveals a close association between
the frequency of price change and product in‡ation, suggesting that price setting behaviour is
mostly driven by local in‡ation in Lesotho. Finally, product price changes at the retail level
within South Africa have an important bearing on price setting behaviour within Lesotho. There
is also a strong evidence that there is market integration in that …rms display similar behaviour
at the product level. This paper therefore makes a number of contributions to the literature.
Firstly, it contributes towards establishing the stylised facts on price setting behaviour within
developing countries. Secondly, the results have implications for the validity of macroeconomic
theory regarding price adjustments and the in‡uence of monetary policy on in‡ation. Finally, the
results provide insight into the extent of product market integration between SA and Lesotho.
Keywords: price setting behaviour, product in‡ation, the frequency of price change
1
1
Introduction
Price setting behaviour forms an important part of economic theory. It plays an important role
in the formulation of monetary policy. It also in‡uences the way in which monetary policy a¤ects
the economy. It has been a great challenge for monetary authorities to improve their knowledge of
how the transmission mechanism of monetary policy works. It may even be more challenging for
the monetary authority of a country which has adopted the monetary policy of another country
such as in the case of Lesotho. For example, if prices are not linked, then it might be the case
that South Africa (SA) monetary policy is not appropriate for Lesotho. Therefore, the knowledge
of price setting behaviour is essential for monetary authorities in Lesotho, particularly in the
adjustment to monetary policy as determined by SA.
Although there is a lot of literature on price setting behaviour, most of research on this area
has primarily been conducted in developed countries. For example, Bils and Klenow (2004) used
individual price data for the U.S; Fabiani et al. (2004) used the survey data collected on Italian
…rms; Alvarez et al (2005, 2006) used consumer price micro data for Spain and the Euro area
respectively.1 Little research has been conducted within the context of developing countries on
price setting behaviour, particularly using disaggregated price level data.2 The main challenge
has been the availability of data at the disaggregated level. This study intends to expand the
scope of this literature in developing countries. Most notably, we use a broad panel of micro-level
price data that is available by product, by retail outlet and across time. With this type of data,
we are able to test if the results conform to the theories of time-dependent and state-dependent
price setting behaviour. We also test for price setting behaviour for Lesotho within the concept of
regional markets.
Analyzing price setting behaviour using micro data o¤ers many insights about the validity and
importance of macroeconomic theory. For example, in‡ation is a macroeconomic variable,which
is underpinned by microeconomic data, and is an aggregate measure of changes in prices at a
unit level. But in‡ation may disguise di¤erent underlying patterns of price changes, such as
price rigidities, which may cause real e¤ects in the economy. Therefore, analyzing prices at a
unit level facilitates an understanding of actual pricing conduct at the most basic level (Creamer
and Rankin, 2008). Studying price setting behaviour in Lesotho will thus add knowledge to the
underlying patterns of price changes such as price rigidities which is particularly relevant for a
country such as Lesotho, an ’island’state within SA.
Classical theories show that most macroeconomic models assume slow price adjustments at
least to generate short-run real e¤ects on monetary policy. The argument is that most prices do
not change daily or even weekly but mostly change infrequently within a year. Their idea is also
1 Other
studies include; Dabusinskas and Randveer (2006) studied price setting behaviour in the Euro area and Estonia; Dyne et al
(2006) studied the Euro area and the U.S.; Dias et al (2004) studied Portugal; and Fisher et al (2000) used Canadian data.
2 Current studies include; Creamer and Rankin (2008) for SA; Kovanen (2006) for Sierra Leone; Julio and Zarate (2008) for Colombia;
Gouvea (2007) for Brazil.
2
that price changes are costless and instantaneous.
Recent theoretical literature show that money and prices can a¤ect the real variables, at least
in the short run due to price stickiness. Price stickiness is usually common in markets with highly
heterogeneous products such as consumer goods markets (Sims, 1998). Individual …rms cannot just
adjust prices because price changes are costly. This explains why setting prices occurs infrequently
or only in response to certain changes in market conditions.
The theoretical literature considers two main forms of price setting behaviour; time-dependent
and state-dependent pricing rules. In time-dependent pricing rule, price changes are a function
of time. Because of costs related to price changes, retailers review their prices at speci…c dates
or discrete time intervals since price reviews and adjustments are costly. A retailer would set
its price in a deterministic manner according to Taylor (1980) or randomly according to Calvo
(1983).3 These models re‡ect exogenous staggering of price changes across …rms in the economy
and therefore a …xed fraction of …rms adjusting their prices in each period (Klenow and Kryvtsov,
2008).4 There is no selection as to who changes their price in a given period.
On the other hand, in state-dependent pricing rule, price changes are a function of the state
in the economy. The main assumption is that there is no routine in price setting; retailers will
change their prices only when there is a speci…c event (for example, a shock). Price changes may
be grouped or spread out, depending on the importance of common against idiosyncratic shocks
and other factors. Prices will therefore be …xed until there is su¢ ciently large shift in market
conditions to warrant a change. Those who change prices are those who will gain most from
doing so. A positive monetary shock increases the fraction of …rms that change prices and /or the
average size of those price changes. An obvious justi…cation for this behaviour is the existence of
…xed costs of changing prices.
Understanding price setting behaviour is important since monetary policies target prices. It
is also important to understand which of the assumptions that underlie macro-models are most
sensible and also help in calibrating these models, which are used by monetary authorities, among
others to set monetary policy. Very little is known about how price setting behaviour in a small
country is linked to that of a neighbouring bigger country. This is especially the case with Lesotho.
Given the foregoing, the subject of this study is centred around the following research questions.
What are the main stylized facts that characterise price setting behaviour in Lesotho? What
explains price setting behaviour in Lesotho? How much is explained by external factors such as
in‡ation in SA?
3A
…xed time interval between two price changes is usually common for products with regulated prices such as education services in
the case of Lesotho.
4 These models allow for the discontinuous price adjustment although they assume that retailers are unable to adjust their prices to
any shock between price adjustment dates.
3
2
The Data
This paper draws on a unique data of highly disaggregated micro-level prices underlying the
consumer price index (CPI) in Lesotho. The distinctiveness of this data is that it is available by
product, retail outlet and location.5 The Bureau of Statistics Lesotho (BOS) collects the product
prices every month from various retail outlets across the 10 districts of Lesotho. It uses a direct
approach to collect price data whereby enumerators physically pay visits to the selected retail
stores in each district. Each individual price record for an item has information on the date
(month and year), retail outlet, district, product category, unit codes and the price of that item.
This approach therefore makes it possible for pricing history of individual items in speci…c retail
outlets to be traced over a long period of time. Using this type of data also has the advantage of
explicitly controlling for aggregation e¤ects that can impact on the convergence of the estimates.
Below are the characteristics of the data used in this study.
Table 1: Number of retail outlets per District
Number of outlets
District
rural
urban
Maseru
14
78
ButhaButhe
0
24
Leribe
1
35
Berea
1
32
Mafeteng
1
27
MohalesHoek
1
28
Quthing
2
25
QachasNeck
0
26
Mokhotlong
4
24
ThabaTseka
4
18
Total
28
317
Total
92
24
36
33
28
29
27
26
28
22
345
Table 1 displays the total number of outlets across districts that are in the data. In total,
there are 345 retail outlets; 317 of which are located across urban centers while 28 are in the
rural parts of the di¤erent districts of Lesotho. Most sampled urban and rural outlets are located
in the Maseru district.6 The outlets consist of grocery stores, auto dealerships, furniture stores,
butcheries, hair salons and others.7
The products are grouped into di¤erent categories as shown in table 2 and into relatively more
disaggregated categories as shown in table 3.
5 By
location we mean rural and urban and also across the ten districts of Lesotho.
is the Capital city of Lesotho and therefore the biggest of all the districts in terms of commercial activities and geographical
6 Maseru
size.
7 the detailed list of outlets used in the sample is in the appendix.
4
Table 2: Number of products by product category
Product category Sub- category
Number of products
Food
41
Perishable
Non-perishable 50
Non-food
Durable
59
Non-durable
56
Services
23
Total
229
In total, there are 229 products in the sample; 40 percent of which are food items and 60 percent
are non-food items. Food items are divided into perishable (18 percent) and non-perishable (22
percent) while, non-food items comprise of durable goods (26 percent), non-durable goods (24
percent) and services (10 percent).
TABLE 3: PRICE RECORDS BY PRODUCT CATEGORY
Number of
Productclass
price records
food
182,090
non-alcoholic beverages
21,062
alcoholic beverages
4,836
tobacco and narcotics
9,094
clothing and footwear
34,407
fuel
20,401
household furniture and equipment
29,644
household operations
24,301
medical care and health expenses
4,298
transport equipment
786
transport related services
663
communications
87
recreation and culture
5,262
education
2,249
personal care
14,758
other goods and services
12,827
Total
366,765
Number of
products
80
8
5
3
31
13
39
8
11
3
2
1
7
2
6
10
229
Percent
49.65
5.74
1.32
2.48
9.38
5.56
8.08
6.63
1.17
0.21
0.18
0.02
1.43
0.61
4.02
3.5
100
Percent
34.07
3.54
2.21
1.33
13.72
5.75
17.26
3.54
4.87
1.33
0.88
0.44
3.1
0.88
2.65
4.42
100
Weighting
to CPI
37.0
1.1
1.0
2.1
17.4
5.6
4.4
4.5
1.1
2.1
6.4
1.2
2.4
2.8
3.2
2.6
94.9
duration
2.5
2.8
3.8
2.6
2.8
2.5
3.7
2.7
4.2
3.1
6.4
4.0
10.3
3.3
3.0
3.6
A further disaggregation of product groups into 16 sub-categories is shown in table 3. In total,
there are around 366, 765 price observations used in this study; 50 percent of which are within
the food category. In comparison to the CPI weighting across di¤erent product categories, speci…c
price data sampled in this study show that a higher proportion is on food,clothing and footwear,
transport related services, fuel and household furniture and equipment.
In terms of the duration of price change, prices of food products change at least once in
every 2.5 months while services exhibit a much larger rigidity. For example, medical services (4.2
months), transport services (6.4 months) and education(10.3 months). This kind of data allows
for a more direct test of theories of price setting behaviour and makes it possible to estimate
the long-run levels of price di¤erentials in di¤erent locations within a country. The speci…city of
5
prices by product de…nition enhances price comparability and minimizes the aggregation biases in
explaining price setting behaviour across product categories and geographic locations.
outliers
Although the BOS requires its price collectors to explain large price changes to reduce measurement errors, some price changes were implausibly large. Therefore, we excluded 1.5 percent
of prices in the price data. We also excluded price data for outlets that were in operation for less
than six months which constituted less than 1 percent of all observations.
3
Facts about in‡ation and price setting behaviour in Lesotho
Lesotho is a member of a customs union and a monetary union.8 The arrangement of the Common
Monetary Area (CMA) was …rst signed as the Rand Monetary Area (RMA) treaty in 5 December
1974 and replaced by CMA in April 1986. The CMA is such that virtually all monetary policy
decisions are determined by the South African Reserve Bank (SARB).9 This implies that Lesotho
essentially adopts monetary policy of SA. The Central Bank of Lesotho (CBL) is mandated primarily to achieve and maintain price stability, but it does not have a domestic in‡ation target. It
has to maintain certain amount of reserves to maintain price stability and the parity between the
Loti and the Rand.
The bene…ts provided by provisions within the CMA and the fact that Lesotho is highly
open have enhanced intra-regional cross border trade between SA and the other three countries
(Lesotho, Namibia and Swaziland). As a result, Lesotho imports about 80 percent of its consumer
goods from SA. Figure 1 below shows the relationship between price movements within the CMA
countries over the past decade.
Figure 1: Inflation rates for CMA countries
8 The
customs union is the Southern African Customs Union (SACU) while the monetary union is the Common Monetary Area
(CMA)
9 under CMA arrangement, Lesotho has pegged its domestic currency to the SA Currency (Rand). Domestic interest rates also follow
those in SA. CBL can only use Open Market Operations (OMO) to control money supply.
6
The diagram shows that there were periods of low in‡ation and periods of high in‡ation in
the past decade across all CMA countries. In Lesotho, during 2002, in‡ation increased to 10.5
percent and went down to 5 and increased to 10.8 in 2008 after which, it started declining again.
Similarly, in‡ation in SA increased to 9.1 percent in 2002, fell to as low as 1.4 percent in 2004
and increased again to reach a peak of 11.5 in 2008. The co-movement in in‡ation rates between
Lesotho and SA can have implications for price setting behaviour in Lesotho. The relationship in
aggregate in‡ation suggests a high degree of integration. But what is driving this relationship is
obscured to explain price setting behaviour in Lesotho. It could be that price in‡ation is di¤erent
across products, or it could be because of asymmetry in price setting behaviour. This implies
that price movements can also be driven by frequency of price change by retailers as well as the
size of price change. Hence why there is need to look at product and retail level data. Aggregate
in‡ation is a weighted average of product level in‡ation. Product level in‡ation, in turn, is the
average of sub-product in‡ation which, on the other hand, is average of in‡ation across retail
outlets, which is related to the frequency of price change and the size of price change at the retail
level. Therefore, our framework is at product-outlet level.10 This is because estimation at a more
aggregate level could lead to downward biased estimates of the median duration of price spells as
a result of heterogeneity at store level (Dias et al. 2004).
4
Review of related literature
The use of micro data in recent years, where studies have examined individual retail price level
data, have o¤ered many insights on the importance of price setting behaviour. In particular,empirical literature on price setting behaviour leads to a better understanding of the frequency
of price change and the determinants of the frequency of price change. Romer (2006) discussed
two main results emanating from this literature.11 The …rst is on the degree of nominal price
rigidity. He showed that prices are not necessarily ‡exible. Blinder (1998) focused on intermediate
goods and services and found that the average frequency of price changes was about a year while
Bils and Klenow (2004) and Klenow and Kryvtsov (2008) used micro data underlying CPI and
found that prices changed in four to six months in the U.S. Dhyne et al. (2005) and Alvarez et al.
(2005) found that the frequency of price change is lower in the Euro area (15.1 percent) than in
the U.S. (24.8 percent).
The second result is that there is a substantial heterogeneity in terms of the frequency of price
change across retailers and products.12 Products di¤er signi…cantly in how frequently their prices
change. At one extreme are products such as fresh food, energy and airfares, whose prices change
at least once every month. At the other extreme are services whose prices change at least once
1 0 Similar
1 1 Klenow
to that which was also used by Baharad and Eden (2004) and by Dias et al (2004)
and Malin (2009) also summarised stylised features of price setting behaviour and the factors that contributed to frequency
of price changes
1 2 This was result was also found by Klenow and Malin (2009) as well as Gopinath and Rigobon (2008)
7
a year. Unprocessed products such as fresh food are usually subject to costs of distribution and
storage. Thus retailers are likely to pass these costs to consumers more quickly to avoid pricing
their products below their marginal costs.13 More cyclical products, particularly transport and
clothing have higher frequency of price change than less cyclical products such as medical care.
In addition, durables appeared to change prices more frequently than non-durables and services,
(Bils and Klenow, 2004 and Klenow and Malin, 2010).
The other …nding from the literature is that there is no evidence about the general downward
price rigidity for all countries, that is, price decreases are only slightly less frequent than price
increases. In addition, price reductions and price increases are of similar magnitude although
on average, price decreases are larger. For example, Benkovskis et al. (2011) found that price
reductions were 10 percent in the Euro area and 14.1 percent in the US while price increases were
8.2 percent in the Euro Area and 12.7 percent in the US.
Researchers have also investigated factors that a¤ect the frequency of price change. Some
studies have found that the level and variability of in‡ation can a¤ect the frequency of price change.
While others point to the frequency and magnitude of cost and demand shocks, the structure and
degree of market competition and changes in the tax rate, sector variables and seasonal e¤ects
among others. A number of empirical studies have analyzed the relationship between price setting
behaviour and in‡ation. The general conclusion of this literature is that various measures of
price setting behaviour are positively related to some aspects of in‡ation. Lunnemann and Matha
(2005) found that also in Luxembourg, around 0.5 percent of the frequency of price change was
explained by accumulated in‡ation, while much higher …gures (18 percent) were found for Austria
in Baumgartner et al. (2005). Aucremanne and Dhyne (2005) found the same results for Belgium,
while Benkovskis et al. (2011) also found that the price decisions of Latvian …rms were mostly
driven by in‡ation at di¤erent levels (overall and product group level). Other studies include,
Vinning and Elwertowski (1976); Parks (1978); Domberger (1987); Van Hoomissen (1988); Lach
and Tsiddon (1992); Debelle and Lamont (1997) and; Loy and Weaver (1998).
In conclusion, the literature shows that prices are not necessarily ‡exible and that there is substantial heterogeneity in CPI disaggregated data across products and across countries. However,
little research has been done in the case of developing countries and much less in the case of Africa
using disaggregated price data. In this study, we …ll this gap by using disaggregated data that
not only vary across products and time, but also across retail outlets and locations (by rural and
urban). This data therefore makes it possible to analyse various aspects underlying price setting
behaviour in Lesotho at the most basic level.
1 3 They
are pro…t maximisers and would not want to price their products below their marginal costs
8
5
Methodological framework
In this section, we outline the di¤erent measures that are used to analyse the stylized facts that
characterize price setting behaviour by products across di¤erent retail outlets in various locations
of Lesotho. There are several measures used to analyse price setting behaviour. These include
the frequency of price change, price decrease and price increase. This paper uses the frequency of
price change as the main measure of price setting behaviour.
5.1
Sample de…nitions
In this context an observation is a quote of a price of an elementary product, that is, a particular
product or a service that is sold in a group of retail outlets, for example, maize meal. A product
category then becomes a representation of elementary products that belong to the same broad
category, for example, food. Earlier in this paper, it was mentioned that the enumerators from
BOS collect the prices of the elementary products periodically (monthly) to develop a series of
price quotes over time.
To assess the changes in the frequency of price changes, the duration of a price spell needs
to be estimated. In this study, we use the frequency approach to determine the frequency of
price changes and the duration of price spells.14 This approach …rst computes the frequency of
price changes as the proportion of times a price for a product in a retail outlet is changed over T
observation periods and then derives the measure of "implied" duration of price spells. The price
spell is referred to as the time interval between two price changes. The duration of the price spell
is then the number of months between the two price changes.15
5.2
Measurement of the frequency of price change
To calculate the frequency of price change, we …rst create an indicator variable that accounts for
price change as:
8
< 1 if p
ijk;t 6= pijk;t
Xijk;t =
: 0 if p
ijk;t = pijk;t
i = 1; :::; I; j = 1; :::::; J; k = 1; ::::; K ; t = 1; ::::; T
1
1
9
=
;
(1)
where pijkt = log price in outlet i, in district j , of product k, in month t
1 4 Other
studies (for example, Benkovskis, 2010 Álvarez, Burriel, and Hernando, 2005 and Gouvea, 2007) used also the duration
approach. However, the drawback of this approach is that it is restricted to work with the uncencored spells only. But the exclusion
of censored spells leads to downward biaslong-lasting spells are more likely to be discarded.
1 5 This approach has its caveat (as emphasised by Baharad and Eden 2004, Baudry and Tarrieu 2004 and Govea 2007), in that it
calculates the inverse of the average frequency of price change instead of the average of the inverse of the frequency of price change. This
causes the former measure to be smaller or equal to the latter due to Jensen’s inequality. However, the authors that have calculated
both measures arrived to a similar order of magnitude for the downward bias (Govea, 2007).
9
We then use the indicator variable to compute the frequency of price change for product k in
retail outlet i in district j over the full period T as:
F req ijk =
1
T
T
X
Xijk;t
for all i = 1; :::::; I; j = 1; :::::; J; k = 1; ::::; K; t = 1; ::::; T
(2)
t=1
This is the frequency of price change for speci…c product sold at a speci…c store. The fraction
of retail outlets that changed the price of product k in district j at time t is calculated as:
F req jk;t =
Ij
X
Xijk;t
for all outlets i = 1; :::::; I
(3)
i=1
where Ij
I , that is a subset of all retail outlets that are located in district j .
The advantage of this method is that it measures the frequency of price change at the product
level for which the successive price observations within a retail outlet are used.16
5.3
Measurement of duration of price spells
The measures for the average duration of price spells are used to estimate the average length of
period between the two price changes. We therefore calculate the average duration of price spells
using the frequency approach as:
durationk =
1
Fk
f or all k = 1; ::::::::; K
(4)
where Fk is the average using frequency of price change for product k at the retail level:
Fk = mean(F reqijk ) over ij or
T
XXX
j
i
i=1
Xijk;t =
1
N
(5)
where N = total observations of ijk for k
5.4
Measurement of the direction of price change
Here, we compute the proportion of times that the price changes were positive or negative throughout the sample for each sector:
8
< 1 if p
ijk;t > pijk;t
Dijk;t =
: 0 if p
ijk;t < pijk;t
We ignore cases where Pijk;t = P ijk;t
1 6 We
1
1
1
9
which indicates a price increase =
which indicates a price decrease ;
(6)
because it means there was no price change.
are calculating two di¤erent types of frequencies . The …rst is at the product*outlet level where aggregation is over time and
it is time invariant. The second is at the product level where aggregation is across outlets and this measure di¤ers with time.
10
6
Empirical Analysis
In this section, we present and discuss some preliminary results from the descriptive statistics to
outline the main stylised facts that characterize price setting behaviour in Lesotho.
6.1
The frequency of price change
As can be observed in …gure 2 that plots the average frequency of price change for the period
2002-2009, there is substantial heterogeneity in the frequency of price change.
Figure 2: Average frequency of price change
The diagram shows that there is substantial heterogeneity in the frequency of price change
among di¤erent products and retail outlets in Lesotho. The estimated median frequency of price
change between the two concurrent months at the product-outlet level is 36 percent. By mean
duration, this implies that more than 50 percent of products change prices in 2.7 months.
0
1
2
3
4
Figure 3:Density of the frequency of price change by product type
0
.2
.4
.6
frequency of price change
food
.8
1
non-food
Band width in both cases is epanechnikov
Figure 3 above shows that there also heterogeneity among food and non-food products. Food
products have a higher frequency of price changes than non-food prices, showing a wider price
dispersion compared to non-food products. On average, 39% of prices of food products change
11
every month while 33% of prices of non-food products change per month. In terms of duration,
prices of food take around 2.6 months to change while non-food products take 3 months. The
‡exibility of food prices could be in‡uenced by a number of factors; supply side constraints such as
weather conditions, input prices, which can a¤ect agricultural productivity (particularly of maize);
high distribution costs due to underdeveloped road infrastructure; and seasonality factors.
0
1
2
3
4
Figure 4:Density of the frequency of price change by product category
0
.2
.4
.6
frequency of price change
perishables
durables
services
.8
non-perishables
non-durables
Band width in both cases is epanechnikov
Disaggregating the product groups further into …ve categories as shown in …gure 4, we observe
that heterogeneity is still substantial. The density functions in general show wide dispersion,
suggesting lots of heterogeneity even within products. Among food products, the frequency of
price change is higher for perishables (44 percent) than for other products. On the other hand,
only 28 percent of services change prices every month. In other words, it takes longer (3.5 months)
for prices of services to change than for prices of other products. This is evidenced by the wider
density function of services which suggests even more substantial heterogeneity within services.
Figure 4(ii):Density of the frequency of price change by product category
0
0
1
1
2
2
3
3
4
4
Figure 4(i):Density of the frequency of price change by product category
0
.2
.4
.6
frequency of price change
perishables
.8
1
0
non-perishables
.2
.4
.6
frequency of price change
durables
Band width in both cases is epanechnikov
.8
non-durables
Band width in both cases is epanechnikov
This result conforms to evidence from the literature that prices for services are generally sticky
(Klenow and Malin, 2010). The stickiness of services may be caused by the fact that services
12
are mostly non-tradable and thus are not subject to high transport and distribution costs. The
lower frequency of price changes for services could also re‡ect the lower volatility of consumer
demand for them. Hence why it is easier for retailers to absorb the increases in the production
costs of services, at least in the short run, to avoid frequent price changes. On the other hand,
prices of perishable food products are relatively more ‡exible compared to other food products.
Unprocessed products such as fresh food are normally subject to relatively high distribution and
storage costs which force retailers to pass these costs to consumers more quickly to avoid pricing
below the marginal cost (Klenow and Malin, 2010). This assumes that distribution costs are more
volatile than other costs because a retailer cannot keep these products for long but need to sell
quickly to get rid of them before they lose durability. Figure 4a con…rms the fact that perishables
have been more volatile than non-perishables even across the entire sample period. This result is
consistent with the results found in Dias et al. (2004) who showed that almost 50 percent of prices
changed for unprocessed food products every month, exceeding that observed for the remainder of
the components of CPI. On the other hand, Coricelli and Horvath (2010) also found that products
such as fruits and vegetables displayed higher frequency of price change (60 percent) while services
had the lowest frequency of price change (15 percent).
0
1
2
3
4
Figure 5a:Density of the frequency of price change by product group
0
.2
.4
.6
.8
1
pfreq
goods
services
Band width in both cases is epanechnikov
Figure 5a re-state what has been found in …gure 4 above that there is more price dispersion
within services than within goods. Evidence also show that there is wide dispersion across locations
in Lesotho, suggesting signi…cant heterogeneity within rural and urban areas as shown in …gure
5 below. However, the greater price dispersion is evidenced in rural areas than in urban areas,
proposing that there is lots of heterogeneity in rural areas than there is in urban areas.
13
0
1
2
3
4
Figure 5:Density of the frequency of price change by retaillocation
0
.2
.4
.6
frequency of price change
urban
.8
rural
Band width in both cases is epanechnikov
Evidence also suggest that prices in rural areas change more frequently than prices in urban
areas. On average 44 percent of prices in rural areas change every month while 36 percent of prices
in urban areas change every month. The implied duration therefore show that prices in rural areas
take at 2.3 months to change while prices in urban areas take 2.8 months. The higher frequency
in rural prices could re‡ect the higher distribution costs in the rural areas relative to outlets
located in urban areas. Rural areas in Lesotho are mostly remote with mostly less developed road
infrastructure compared to roads in urban areas. Therefore the costs of distributing products to
the rural areas is relatively high for rural retailers than for urban retailers, thereby forcing them
to review their prices more frequently. The results could also be attributable to the fact that there
are fewer outlets in the rural areas compared to urban areas. Therefore rural consumers have fewer
retailers to choose from hence why it is easier for rural retailers to change prices more frequently
than it is for urban retailers. However, this results must be treated with caution because the
observed heterogeneity might also be due to di¤erences in product composition across locations.
This is particularly because the composition of products approximates food products; 60 percent
of retail outlets sell food products, while in urban areas the composition is represents a wide range
of products.17 In this analysis, we lumped all months together and therefore included periods of
relatively high and low in‡ation. In the next section, we want to look at how price changes di¤er
across time in more detail.
1 7 see
appendix for the detailed table of types of outlets by location.
14
6.2
Frequency of price increases and price decreases
Figure 6: Monthly frequency of price changes
Figure 6 above shows that there is a higher frequency of price increases than price decreases
throughout the sample period. Regarding the frequency of price decreases, there has been almost
constant movement throughout the sample period but showing high volatility. Movements in the
frequency of price increases generally follow changes in the overall in‡ation (in …gure 1). There
seem to be a downward movement from 2002 to 2005 and then an upward movement until 2008
and thereafter, the trend started going down again.
Figure 7: Frequency of price changes by product category
Figure 7 indicates that across all product categories, average frequency of price increases dominates the frequency of price decreases. There is also signi…cant heterogeneity in price setting
behaviour across di¤erent product categories. We note that perishable food prices increase more
frequently than any other product category while prices of services increase less frequently than
other categories. Price decreases follow the similar pattern as the price increases with perishable
products having the highest frequency and services having the lowest frequency. This result reproduces the result shown in …gure 4b that perishable products generally have more frequent price
changes while services have the least frequency in price changes.
15
Figure 8: Frequency of price changes by product group
Figure 8 shows the frequency of price change, increase and decrease in the most disaggregated
groupings. According to the diagram, there is heterogeneity in price setting behaviour among
di¤erent disaggregated product categories. In general, the frequency of price change is fairly high
across all product groups. However, prices of food products seem to change more frequently than
other products while education and related expenses change relatively less frequently. As in …gure
7, the frequency of price increases is greater than the average frequency of price decreases across
all product groups. Prices of tobacco and narcotics increase fairly more frequently while education
and related expenses increase less frequently. On the other hand, education and related expenses
are relatively stickier compared to other product groups. Clothing and footwear prices decrease
more frequently, which seems to be also the case with SA as mentioned in Creamer and Rankin
(2008). They showed that for footwear products, the frequency of price decreases were greater
than frequency of price increases. This could be explained by seasonal factors and end-of month
discount sales which are common in that product category.
7
Comparison with international evidence
Table 4: Comparison with international evidence
Country
Frequency of price change
Mean duration (in months)
Lesotho (2002-09)
36.4
2.7
SA (2001-07)
17.1
4.2
Euro Area (1996-01)
15.1
6.6
USA (1998-03)
26.1
3.8
Spain (1993-01)
15
France (1994-03)
18.9
Slovakia (1997-01)
34
3.8
Brazil (1996-06)
37
2.7
Sierra Leone (1999-03)
51
16
Source: Gouvea (2007); Creamer and Rankin (2008); and Coricelli and Horvath (2010)
Table 4 above shows the comparison of the results found in this paper with results found in
other countries. However, we should interpret these results with caution because of di¤erences in
product composition and duration of years between countries. It was found earlier that on average,
36 percent of prices in Lesotho change every month. By implied mean duration, prices take 2.7
months to change in Lesotho. These results can be compared with the industrialized countries
and other developing countries. In the US, the average frequency of price change was found to be
26.1 percent and the mean duration 3.8 months. Prices in the Euro area are the most rigid; the
average frequency was 15.3 percent and the median duration was 6.6 months.
In terms of developing countries, Sierra Leone had the highest average frequency of price change
of 51 percent mainly due to in‡ation uncertainty (Kovanen, 2006). Prices in Slovakia were also
relatively ‡exible with the frequency of 34 percent while the duration was 3.8 months. Brazil
prices could be directly comparable to Lesotho’s prices which change more frequently than prices
in Slovakia and SA. The frequency of price change in SA was found to be 17.1 percent (Creamer
and Rankin, 2008) which shows that prices in SA are relatively less ‡exible than prices in Lesotho.
In fact, Lesotho’s prices are twice more ‡exible than prices in SA.
In conclusion, the results show that the price setting behaviour exhibits substantial degree
of outlet, product and sector heterogeneity. In the food sector, prices remain unchanged for
a relatively shorter period of time while in the services sector, prices remain unchanged for a
relatively longer period. There is also evidence of heterogeneity across locations which could
be more in‡uenced by product composition than di¤erences due to location. Di¤erences in the
frequency of price change across countries also suggests heterogeneity across countries although
this result could be marginalised by the di¤erences in product composition and duration of period
of study.
8
Determinants of the frequency of price change
This section now investigates the determinants of frequency of price change, focusing particularly
on the role of in‡ation. The frequency of price change is used as a measure of price setting behaviour. The importance of price stickiness has been subject to much recent macroeconomic research.
This is because of the importance of the evolution of the frequency of price change as well as its
determinants in understanding macroeconomic ‡uctuations. According to state-dependent pricing
rule, as argued by Cecchetti (1986), the frequency of price change is almost certainly dependent
on the economic environment. In this context, the interest lies in the relationship between the
relationship between the price in‡ation and the frequency of price change. For example, larger
accumulated in‡ation is associated with higher frequency of price change.
Related to the frequency of price change is the structure of the cost of price change. State-
17
dependent pricing rule assumes that prices are sticky because …rms face costs of changing their
prices,which are assumed to be …xed (in real terms) as in Sheshinski and Weiss (1977).18 When
these costs of changing prices are high, in‡ation a¤ects the frequency of price change. The presence
of …xed costs of adjustment in nominal prices encourages the …rm to change its nominal prices in
a discontinuous manner. When the bene…ts of changing prices are high or the costs are low, then
…rms will change their prices more frequently. Ceccetti (1986) indicated that in the assumption
that price change is …xed, the probability of changing the price depends on various variables
describing the last price change such as in‡ation and demand. That is why the frequency of
price change depends on the state of the economy. When in‡ation is high for example, then the
frequency of price change should be high because in‡ation causes the bene…t of changing prices to
increase overtime (Wolman, 2000).
Macroeconomic theory show that the frequency of price change can also be in‡uenced by other
aspects of in‡ation. There is a relationship between current and expected in‡ation through the
New Keynesian Phillips Curve (NKPC) approach. However, a pure forward-looking approach
seem to be unable to generate enough in‡ation persistence to match the data, (Greenslade and
Parker, 2010). Gali and Gertler (1999) therefore proposed an alternative ’hybrid’Phillips curve
relationship which allows for lagged in‡ation to also a¤ect the frequency of price change. They
argue that while only a certain fraction of …rms choose to set prices at the optimal level each period,
the remainder subset of …rms do not receive the Calvo signal to change prices and therefore use a
backward looking rule of thumb to set prices. This relationship allows for the analysis of in‡ation
persistence on price setting behaviour.19
On the other hand, the approach by Lucas (1973) based on incomplete information indicates
that the inability of …rms to di¤erentiate between aggregate and local shocks causes the di¤erences
in price setting behaviour. They face a trade-o¤ between paying attention to aggregate conditions
and paying attention to idiosyncratic conditions. Under this approach, the frequency of price
change can also be a¤ected by the unexpected in‡ation (Lach and Tsiddon, 1992).20
Given the above, this paper contributes to the literature on the importance of product in‡ation
on …rm’s frequency of price change in the case of Lesotho. Several studies have modi…ed the
Ceccetti (1986) speci…cation. For example, Aucremanne and Dyne (2005) included accumulated
aggregate in‡ation at di¤erent levels of sectors. Benkouskis et al. (2011) included accumulated
consumer price in‡ation at the most basic level (6-digit COICOP), at the 2-digit COICOP level
that describes price changes at the group level and accumulated in‡ation since the last price
1 8 These
costs are categorised into; physical adjustment costs, managerial costs and customer costs. Physical costs involve the actual
implementation of a price change while customer costs consist of the time spend conveying price changes to customers, time spend
negotiating prices with customers and costs associated with loss of sales because of antagonizing customers. Managerial costs are also
known as menu costs because they are usually assumed to be …xed (independent of the magnitude of price change). They include
personnel time in decision making, recording, calculating and posting new prices.
1 9 Amato and Laubach (2003), Steinsson (2003), and Strum (2010) also make the same assumption.
2 0 Lach and Tsiddon (1992) decomposed in‡ation into expected and unexpected in‡ation. Using disaggregated price data, they found
that the e¤ect of expected in‡ation on price setting behaviour is stronger than the e¤ect of unexpected in‡ation
18
change. In this analysis, we use product in‡ation instead of consumer price indices because the
former is superior in terms of producing robust results and therefore better re‡ects local supply
and demand conditions. Further, product in‡ation eliminates possible store e¤ects in price levels.
In addition, given the heterogeneity found in the data, it would not make sense to use aggregate
measure of in‡ation. As mentioned by Cecchetti (1986), individual in‡ation also re‡ects changes
in demand or supply conditions in a given market which a¤ect the optimal price. We measure
individual product in‡ation by mean price changes within product groups.
While the factors that in‡uence the …rm’s behaviour on price setting are critically important
to understanding the workings of the market economy, and thus the consequences of public policy,
particularly monetary policy, empirical evidence on this issue remain scarce especially in developing
countries.
Assuming a state-dependent pricing rule, we test the relationship between the frequency of
price change and product in‡ation as follows:21
f req jk;t =
+
1 dlpjg;t + jk
(7)
Where:
dlpjg;t =
PIj Pkg
i=1
i=1
dlpijk;t = lpjg;t
lpjg;t
1
which is monthly average log of price change for
product category g within outlet i in district j
A positive and statistically signi…cant coe¢ cient of product in‡ation would indicate that …rms
in Lesotho generally follow a state-dependent pricing rule. We use log of price changes at the
product group level in order to be able to analyse whether …rms’decisions are a¤ected by price
changes in a particular market.
Following Gali and Gertler (1999),we introduce in‡ation persistence into the regression. This
is based on the assumption that …rms can be irrational and therefore be backward looking when
setting prices. In other words, price setting behaviour does not only depend on current in‡ation
but also on past values of in‡ation. We therefore include the previous values of quarterly product
in‡ation in the speci…cation below, :
f req jk;t =
+
1 dlpjg;t + 2 lag
_dlpjg;t 1 +
(8)
jk
The previous sections revealed substantial heterogeneity across products and locations. We
therefore include the product and district dummies to control for this unobserved heterogeneity.
Monthly dummies are also included and take into account changes in overall in‡ation and also
capture seasonal e¤ects. The speci…cation in this case is as follows:
f req jk;t =
2 1 Such
+
1 dlpjg;t +
2 lag
_dlpjg;t
measures include the frequency and the size of price change.
19
1
+
k
+
j
+
t + ijk
(9)
The approach by Lucas (1973) proposes that …rms fail to distinguish between local and aggregate in‡ation shocks because of imperfect information, and thus set prices di¤erently. This
approach emphasizes the role of unexpected in‡ation on the frequency of price change. Using the
model of sticky prices, Mackowiak and Wiederholt (2009) showed that …rms pay more attention
to idiosyncratic conditions than to aggregate conditions because idiosyncratic conditions could be
more variable. We therefore include national in‡ation (dlp_national) and the standard deviation
of product in‡ation (sd_dlp) within the group to capture unexpected in‡ation. :
f req ijk;t = +
1 dlpjg;t + 2 lag
_dlpjg;t
1
+
_nationalg;t +
3 dlp
4 sd
_dlpg;t +
k
+
j
+
ijk
(10)
If national in‡ation is signi…cant, it would mean that retailers also pay attention to aggregate
conditions. This also imply that the coe¢ cient of national in‡ation would be capturing marginal
impact attributable to the region. On the other hand, the signi…cance of unexpected in‡ation
would mean that the frequency of price change in Lesotho is a¤ected by unexpected in‡ation.
8.1
Discussion of results
This section presents and discusses the results found on the role of in‡ation dynamics on the
frequency of price change in Lesotho.
Figure 9: Average monthly frequency and the product inflation
Figure 9(a) plots the average monthly in‡ation and the frequency of price change over the
sample period April 2002 to December 2009. The diagram shows that there is a similarity in the
movement of the frequency of price change, product in‡ation and aggregate in‡ation. The diagram
also show that there was a gradual decline in average price changes from 2002 until mid 2005 after
which they started to increase until 2008. This movement follows the trend in the overall in‡ation
rate as shown in the earlier section. A downward trend in the frequency of price change is related
to the decline in in‡ation around the same period,while an upward movement is linked to an
increase in in‡ation. Figure 9(b) on the other hand shows a scatter plot of monthly frequency
of price change against monthly average product in‡ation between April 2002 to December 2009.
20
It gives a visual evidence that there is a positive relationship between product in‡ation and the
frequency of price change in Lesotho. The correlation coe¢ cient between the frequency of price
change and in‡ation is positive and signi…cant as found by Bils and Klenow (2004) as well as
Álvarez (2008). This is also revealed by the signi…cant and positive coe¢ cient of the regression
in column 1 of table 5. To test whether the correlation is robust to inclusion of various controls
and …xed e¤ects, table 5 presents the results of speci…cation (7) and (9) above, which takes into
consideration other factors that might a¤ect the frequency of price change.
Table 5: Estimated Regressions on the Frequency of Price Change and Product Inflation
dlp
lag_dlp
I_dlp_rural
sd_dlp
(1)
Coefficient
0.351***
(0.084)
(2)
Coefficient
0.452***
(0.121)
0.338***
(0.108)
-0.086
(0.256)
-0.007
(0.020)
dlp_national
Dnov
(3)
Coefficient
0.417***
(0.113)
0.310***
(0.097)
-0.043
(0.191)
-0.026
(0.019)
(4)
Coefficient
0.320***
(0.099)
0.310***
(0.096)
0.001
(0.184)
-0.022
(0.018)
0.577**
(0.239)
0.646***
(0.045)
0.615***
0.628***
0.599***
(0.037)
(0.029)
(0.035)
Djan
0.576***
0.583***
(0.030)
(0.024)
Constant
0.290***
0.415***
0.442***
0.438***
(0.005)
(0.014)
(0.024)
(0.023)
R2
0.088
0.215
0.368
0.370
N
12782
11959
11959
11959
Fixed effects
None
Product, District
Product, District, Month
Product, District, Month
Notes: The dependant variable is the average frequency of price change. Regression (1) is the basic regression with only product
inflation as an independent variable. Regression (2) includes other control variables such as rural inflation, lagged inflation and
unexpected inflation (3 includes time effects. Regression (4) includes national inflation and also control for unobserved
heterogeneity. Robust and clustered standard errors are in parenthesis below the estimated coefficients. *** Significant at 1 percent
level ** significant at 5 percent level * significant at 10 percent level.
Evidence from table 5 shows that the coe¢ cients of product in‡ation are positive and signi…cant
at 1 percent in all the regressions. This suggests that Lesotho retailers follow a state-dependent
pricing rule. This also suggests that local in‡ation matter to retailers in Lesotho. The basic
speci…cation with no controls (in the …rst column) indicates that a 1 percent increase in in‡ation
increases the frequency of price change by 0.35 percent. The magnitude of this coe¢ cient approximately displays the mean frequency of price change. To control for other factors that might also
a¤ect the frequency of price change as stated in theory, rural in‡ation (I_dlp_rural), previous
in‡ation (lag_dlp), unexpected in‡ation (sd_dlp), district and product dummies are included
in column 2 . The coe¢ cient of product in‡ation remains signi…cant and positive ( moves from
0.35 to 0.45) showing that the e¤ect of a unit increase in product in‡ation increases frequency of
price change by 0.45. Previous in‡ation (lag_dlp) is also positive and highly signi…cant , showing
positive relationship between in‡ation persistence and the frequency of price change. This also
suggest that there is a long-run relationship between in‡ation and the frequency of price change; 10
percent increase in in‡ation leads to 0.06 to 0.08 increase in the frequency. However, unexpected
in‡ation was found insigni…cant. It seems in the case of Lesotho, unexpected in‡ation does not
21
a¤ect the frequency of price change.
The inclusion of time …xed e¤ects in column 3 controlled for seasonal and other time e¤ects.
The coe¢ cients of product in‡ation and previous in‡ation however remain signi…cant. In the
fourth speci…cation, we included national in‡ation to control for the e¤ect of national conditions
(relative to local conditions) on the frequency of price change. If we do not account for in‡ation
at national level, then the e¤ect of local in‡ation is biased upwards (as indicated by the coe¢ cient
of product in‡ation in column 3) because local in‡ation will be correlated with national in‡ation.
The results, in column 5, show that product group in‡ation at national level has a strong positive
e¤ect on the average frequency of price change, suggesting that retailers in Lesotho pay attention
to local and national conditions when setting their prices.
Table 6: Estimated regressions of frequency of price decrease and increase and product inflation
dlp
Price decrease
Coefficient
-0.672***
(0.079)
lag_dlp
I_dlp_rural
sd_dlp
dlp_national
Dnov
Djan
Constant
R2
N
0.507***
(0.026)
0.144***
(0.013)
0.111***
(0.008)
0.188
12782
Price decrease
Coefficient
-0.752***
(0.071)
0.064
(0.059)
0.362***
(0.128)
-0.011
(0.010)
-0.077
(0.095)
0.520***
(0.028)
0.175***
(0.018)
0.136***
(0.012)
0.386
11959
Price increase
Coefficient
0.981***
(0.150)
0.104***
(0.012)
0.399***
(0.028)
0.175***
(0.015)
0.155
12782
Price increase
Coefficient
1.034***
(0.096)
0.237***
(0.077)
-0.378**
(0.188)
-0.003
(0.018)
0.643***
(0.203)
0.120***
(0.035)
0.313***
(0.037)
0.300***
(0.025)
0.356
11959
Fixed effects
None
Product, District, Month
None
Product, District, Month
Notes: In column (1) and (2), the dependant variable is the average frequency of price decrease and in column (3) and (4) is the
frequency of price increase. Standard errors are in parenthesis below the estimated coefficients are robust and clustered. ***
Significant at 1 percent level ** significant at 5 percent level * significant at 10 percent level.
In addition, we regressed equation (9) against the frequency of price increase and the frequency
of price decrease. The results are similar for the frequency of price increase but are di¤erent for
the frequency of price decrease as depicted in table 6.
In the previous section, the frequency of price change in food products were found to be
relatively higher than in other products. We now test the relationship between the frequency of
price change and in‡ation across all products including food. Table 7 shows sectoral speci…cations
of 15 di¤erent product categories. Column 1 shows the e¤ect of food in‡ation on the frequency
of price change and the results show that product in‡ation positively a¤ects the frequency of
price change by a large magnitude. In fact, the coe¢ cient of food in‡ation is greater than the
coe¢ cients of other product categories. This could imply that in Lesotho, food in‡ation in‡uences
the decisions of retailers to change prices more frequently more than in‡ation of other products.
The results also show that in‡ation in household furniture and equipment; medical care and health
22
services; transport services; recreation and culture do not a¤ect the frequency of price change in
Lesotho. These results show that, in Lesotho, individual product in‡ation a¤ect price setting
behaviour di¤erently across di¤erent sectors, which means that sectoral response to monetary
policy is likely to be di¤erent.
Table 7: Estimated regressions of frequency of price change and inflation by product categories
dlp
lag_dlp
sd_dlp
I_dlp_rural
Djan
food
Non-alcoholic
Alcoholic
Coefficient
1.796***
(0.607)
-2.425**
(1.055)
-0.121
(0.179)
-0.156
(0.797)
0.368***
(0.042)
Coefficient
0.779*
(0.413)
-0.732
(0.762)
-0.179
(0.131)
-0.514
(0.491)
0.515***
(0.094)
0.689***
(0.094)
0.445***
(0.071)
0.530
873
Yes
Coefficient
1.172***
(0.218)
0.462
(0.325)
0.040
(0.052)
-0.381
(0.350)
0.630***
(0.168)
0.785***
(0.175)
0.313*
(0.166)
0.490
849
Yes
Dnov
Constant
0.585***
(0.031)
0.654
873
Yes
R2
N
Fixed effect
Tobacco &
narcotics
Coefficient
1.504***
(0.395)
0.111
(0.714)
-0.059
(0.140)
0.261
(0.521)
0.274**
(0.138)
0.946***
(0.117)
0.538***
(0.077)
0.441
873
Yes
Clothing &
footwear
Coefficient
0.587***
(0.226)
0.076
(0.447)
-0.090
(0.066)
0.333
(0.596)
fuel
h/h furniture &
equipment
Coefficient
0.278
(0.203)
0.460
(0.301)
Coefficient
1.158***
(0.288)
-0.052
(0.565)
-0.312***
(0.093)
0.127
(0.578)
0.571***
(0.080)
0.678***
(0.077)
0.383***
(0.074)
0.521
873
Yes
0.754***
(0.037)
0.086***
(0.029)
0.544
873
Yes
1.366
(0.918)
0.447***
(0.042)
0.461***
(0.042)
0.283***
(0.051)
0.721
812
Yes
h/h
Medical care &
Transport
Transport
Recreation &
Education
Personal care
operations
health services
equipment
services
culture
Coefficient
Coefficient
Coefficient
Coefficient
Coefficient
Coefficient
Coefficient
dlp
0.843**
0.354
0.604**
0.010
-0.028
0.300**
0.604***
(0.360)
(0.215)
(0.300)
(0.161)
(0.184)
(0.141)
(0.225)
lag_dlp
-0.104
0.309
-0.125
0.185
-0.555**
0.060
-0.347
(0.669)
(0.190)
(0.615)
(0.214)
(0.253)
(0.174)
(0.429)
sd_dlp
-0.041
-0.000
-0.058
-0.002
-0.021
0.015
-0.165**
(0.107)
(0.029)
(0.113)
(0.027)
(0.048)
(0.027)
(0.073)
I_dlp_rural
-0.108
-0.251
0.153
10.898
-0.573
(0.455)
(0.251)
(0.430)
(14.601)
(0.455)
Dnov
0.575***
0.601***
-0.128
0.204
0.865***
0.545***
0.901***
(0.074)
(0.108)
(0.212)
(0.205)
(0.072)
(0.103)
(0.050)
Djan
0.381***
0.556***
0.034
0.217
0.846***
0.518***
0.855***
(0.079)
(0.090)
(0.063)
(0.224)
(0.045)
(0.111)
(0.041)
Constant
0.662***
0.290***
0.896***
0.898***
0.017
0.003
0.127***
(0.062)
(0.074)
(0.055)
(0.214)
(0.029)
(0.044)
(0.034)
R2
0.543
0.438
0.419
0.626
0.504
0.784
0.576
N
877
851
475
496
811
596
873
Fixed effects
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Notes: The dependant variable is the frequency of price change. The fixed effects are for products, districts and month.
*** Significant at 1 percent level ** significant at 5 percent level * significant at 10 percent level.
8.1.1
Other goods &
services
Coefficient
0.547*
(0.283)
-0.072
(0.406)
-0.035
(0.097)
-0.395
(0.506)
0.703***
(0.111)
0.674***
(0.108)
0.317***
(0.109)
0.501
877
Yes
The role of SA price dynamics on price setting behaviour in Lesotho
Earlier, we illustrated the co-movement of the SA and Lesotho in‡ation rates. It is therefore
interesting to test whether price dynamics in SA could in‡uence price setting behaviour in Lesotho.
In this section, we analyse the role that in‡ation dynamics in SA play on price setting behaviour
in Lesotho. This will also capture the e¤ect of in‡ation at the regional level rather than just
the national level as in the previous section. We therefore augment equation (9) to capture the
regional conditions. The speci…cation in this case is as follows:
f req k;t
=
+
+
1 dlp
_Lesk;t +
6 f reqsak;t
+
2 dlp
_Lesk;t
6 f reqsak;t 1
1
+
3 dlp
_SAk;t +
+ dlp_nationalk;t +
_SAk;t
4 dlp
k
+
t+ k
Where:
f reqsak;t = average frequency of price change in SA in period t.
f reqsak;t
1
= average frequency of price change in SA in period t - 1
dlp_Lesk;t = product in‡ation in Lesotho in period t
23
1
+
5 dlp
_SAk;t
2
(11)
dlp_Lesk;t
1
= product in‡ation in Lesotho in the period t - 1
dlp_SAk;t = product in‡ation t in SA in period t
dlp_SAk;t
1
= product in‡ation in SA in the period t - 1
dlp_SAk;t
2
= product in‡ation in SA in the period t - 2
Table 7 presents the estimated results on the e¤ect of in‡ation dynamics between Lesotho and
SA on the frequency of price change in Lesotho. We mapped similar products between Lesotho
and SA price data which came to 132 products and 6,858 observations in total.
Table 7: Estimated regressions of frequency of price change and inflation between Lesotho and SA
(1)
Coefficient
0.206***
(0.074)
(2)
perishables
non-perishables
non-perishables
Coefficient
Coefficient
Coefficient
Coefficient
dlp_Les
0.223***
0.062
0.275***
(0.080)
(0.140)
(0.099)
dlp_Les (t-1)
0.182
0.021
-0.229
(0.154)
(0.258)
(0.282)
dlp_SA
0.059
-0.019
-0.084
0.468**
0.236
(0.126)
(0.125)
(0.188)
(0.183)
(0.184)
dlp_SA (t-1)
0.075
-0.069
0.638***
0.376**
(0.093)
(0.175)
(0.205)
(0.185)
dlp_SA (t-2)
0.142
0.039
0.468**
0.336*
(0.129)
(0.176)
(0.192)
(0.183)
freqsa
0.133***
0.210***
0.112***
0.125***
(0.027)
(0.057)
(0.030)
(0.030)
freqsa (t-1)
0.100***
-0.016
0.063*
0.072**
(0.026)
(0.054)
(0.033)
(0.033)
dlp_national
0.221
0.456
0.531
(0.255)
(0.406)
(0.557)
Dnov
0.550***
0.559***
0.500***
0.448***
0.597***
(0.037)
(0.030)
(0.112)
(0.042)
(0.052)
Djan
0.538***
0.528***
0.498***
0.387***
0.557***
(0.030)
(0.026)
(0.112)
(0.030)
(0.045)
Constant
0.390***
0.359***
0.397***
0.504***
0.343***
(0.027)
(0.024)
(0.110)
(0.025)
(0.037)
R2
0.433
0.437
0.515
0.614
0.590
N
6631
5997
1378
2000
2096
Fixed effects
Product, month
Product, month
Product, month
Product, month
Product, month
Notes: The dependent variable is the average frequency of price change. The second regression of non-perishables excludes
Lesotho inflation and its previous values. *** Significant at 1 percent level ** Significant at 5 percent level * significant at 10
percent level. This is a within product variation regression.
In table 7, estimations in the …rst two columns used the whole sample of products. The results
show that there is a positive signi…cant relationship between the frequency of price change and
product in‡ation in Lesotho. The frequency of price change in SA is also positive and signi…cant,
indicating that price setting behaviour in SA a¤ects the way in which retailers in Lesotho set their
prices. This suggests the relationship in the ownership of …rms in Lesotho and in SA. This is
because most dominating retail outlets in Lesotho are owned by SA citizens who also own outlets
in SA and this can therefore have an in‡uence on the overall behaviour of retailers in Lesotho.
The results suggests that not only local markets matter for retailers in Lesotho but also regional
markets. This also points out to the importance of regional integration in price setting behaviour
in Lesotho. The past values of the frequency of price change in SA are also signi…cant, re‡ecting
the fact that retailers in Lesotho also consider the history of price setting behaviour in SA when
setting their prices.
24
In column (3),(4) and (5), we considered only perishable and non-perishable products respectively. The results show that for perishable products, only the frequency of price change in SA is
related to the frequency of price change in Lesotho. This means that retailers in Lesotho and SA
have common underlying price setting behaviour for perishable products. This is also expected for
integrated markets. However, for non-perishable products, only the lagged values of in‡ation in
SA are signi…cant and positive. This indicates that retailers in Lesotho consider only the history
of in‡ation in SA when setting their prices for non-perishable products, suggesting that it might
take longer for the e¤ects of SA in‡ation to …lter through into Lesotho. The results also indicate
that there is no relationship between in‡ation in SA and the frequency of price change in Lesotho.
The frequency of price change in SA is also related to the frequency of price change in Lesotho.
These results are expected because non-perishable products are highly tradable between Lesotho
and SA and can be stored. The results for other non-food products show insigni…cant coe¢ cients
as indicated in the appendix. The other important result is that in this case, national in‡ation is
not signi…cant. This could suggest that in the previous section (table 5), national in‡ation was
marginally capturing regional e¤ects.
8.2
Further analysis: the frequency and the size of price change
In the presence of monopolistic competition where substitute goods are readily available, the costs
may be proportional to the size of the real price change or the decreasing function of the frequency
of price change, (Ceccetti, 1986).Therefore it is important to look at the relationship of the frequency and the size of price change assuming that price changes are costly. It is also important to
look at the distribution of the size of price changes because of the substantial heterogeneity found
in the data.
Table 9: Size and Frequency of price changes
PRODUCT CLASS
food
non-alcoholic beverages
alcoholic beverages
tobacco and narcotics
clothing and footwear
fuel
household furniture and equipment
household operations
medical care and health expenses
transport equipment
transport related services
recreation and culture
education
personal care
other goods and services
SIZE OF PRICE CHANGE
Price
Price
Price
change
increase
decrease
1.6%
8.5%
-10.2%
1.4%
8.2%
-9.7%
3.4%
11.4%
-11.8%
2.0%
6.1%
-7.3%
0.9%
12.9%
-15.0%
1.8%
8.6%
-10.2%
1.1%
18.3%
-21.7%
1.5%
8.8%
-10.2%
2.5%
17.4%
-20.3%
0.2%
8.2%
-12.8%
1.5%
20.6%
-33.2%
0.8%
17.9%
-20.2%
5.7%
34.4%
-35.9%
1.7%
11.7%
-14.1%
1.8%
11.1%
-13.5%
25
FREQUENCY OF PRICE CHANGE
Price
Price
Price
change
increase
decrease
38.1%
24.0%
14.1%
34.7%
21.6%
13.1%
25.4%
16.6%
8.8%
37.6%
26.1%
11.5%
35.0%
20.1%
15.0%
38.3%
24.4%
13.9%
25.6%
14.6%
11.0%
36.1%
22.2%
13.9%
23.0%
13.9%
9.1%
31.0%
19.1%
11.9%
14.6%
9.4%
5.2%
24.0%
13.2%
10.8%
9.4%
5.6%
3.8%
29.1%
17.9%
11.3%
31.9%
19.9%
12.1%
Table 9 shows that on average, price changes are small in size but occur more frequently.
Price increases occur more frequently than price decreases and their size is systematically smaller
than those of price decreases. This indicates that the product categories whose prices change
relatively more frequently, change by a relatively small magnitude and vice versa. For example,
food products are fairly ‡exible (38%) compared to the rest of the products but the magnitude by
which their prices change is relatively small (1.6%). On the other hand, education and transport
services seem to be less ‡exible (9.4%) and (14.6%) with the size of price change of 5.7% and
1.5% respectively. This re‡ects a negative relationship between the size and the frequency of price
change across all product groups, indicating asymmetry in pricing conduct.
The general theory of …rm price setting behaviour as in Sheshinski and Weiss (1977) shows
that there is a link between the asymmetry in pricing conduct and the cost of changing prices.22
This theory predicts that when setting its prices, the …rm compares the total bene…t with the
total cost of changing the price and changes the price when the net bene…t is positive. Large costs
of adjusting the price reduce the net bene…t of changing that price. This implies that, holding
everything constant, …rms with a higher costs of changing prices will infrequently make larger
price changes. Firms that change their prices less frequently need to make large changes to bring
back prices to the desired level when the change is …nally observed.
These results seem to be consistent with hypothesis indicated by Sheshinski and Weiss (1977)
that high frequency of price change leads to a small size of the price change and vice versa.
Evidence also show that sizable price increases tend to display sizable price reductions. For
example, education has a sizable price increase (34.4 percent) as well as a sizable price decrease
(35.9 percent), displaying downward price rigidity.
In order to statistically check for the consistency of this hypothesis, we estimated the following
descriptive relationship:
sizedlpk;t =
+
1 f req k;t + ij;k
(12)
where;
sizedlpk;t is the average size of price change for product k at month t
f req k;t is the average frequency of price change for product k at month t
We also estimated price increase and price decrease to check for consistency of our results
above:
dlp_plusk;t =
+
1 f req
_plusk;t +
dlp_ min usk;t =
+
1 f req
_ min usk;t +
where;
2 2 This
was also indicated in Powers and Powers (2001)
26
ij;k
ij;k
(13)
(14)
dlp_plusk;t is the average size of price increase for product k at month t
f req _plusk;t is the average frequency of price increase for product k at month t
dlp_ min usk;t is the average size of price decrease for product k at month t
f req _ min usk;t is the average frequency of price decrease for product k at month t
In all the three speci…cations, we controlled for unobserved product characteristics and monthly
…xed e¤ects. The results are presented in table 10:
Table 10: Estimated regressions of frequency and size of price change, increase and decrease
(1)
(2)
(3)
The size of price change
The size of price increase
The size of price decrease
Coefficient
Coefficient
Coefficient
freq
-0.013*
(0.008)
freq_plus
-0.071***
(0.007)
freq_minus
0.052***
(0.010)
Dnov
-0.108***
0.143***
-0.157***
(0.014)
(0.019)
(0.019)
Djan
0.023*
0.122***
-0.176***
(0.012)
(0.012)
(0.020)
Constant
0.052***
0.101***
-0.050***
(0.004)
(0.009)
(0.013)
R2
0.047
0.318
0.295
N
15357
13544
11713
Fixed effects
Product, month
Product, month
Product, month
Notes: *** Significant at 1 percent level ** significant at 5 percent level * significant at 10 percent level.
This is a within product variation regression.
The table shows a signi…cant negative relationship between the frequency of price change and
the size of price change. The …rst column shows that the size of price change is negatively related
to the frequency of price change as we saw in table 8 above, although the signi…cance is marginal.
The table also shows that there is a strong negative relationship between the frequency of price
increase and the size of price increase. Firms are usually hesitant to increase prices than to reduce
them due to related costs. But when the changes occur, they are usually signi…cant to cover for
the losses incurred for not changing the prices. These results are in line with the relationship
displayed in table 9 and are consistent with the hypothesis by Sheshinski and Weiss (1977) of
asymmetric price setting behaviour. These results suggest that there are related costs to changing
prices which exceed the bene…t causing …rms to infrequently change their prices and make large
changes when they do so.
9
Conclusion
The main objectives of this study were; to document price setting behaviour and compare this to
the existing literature; to investigate if there is evidence of time or state-dependence in Lesotho;
and to examine whether price setting behaviour in Lesotho is correlated with SA price setting
behaviour. The following conclusions were drawn from the results.
27
The frequency of price change as determined by the total number of non-zero price change
observations o¤ers some evidence on the degree of price rigidity. The data indicate that on average,
36% of prices change in a month in Lesotho. Correspondingly, 64 percent of prices remain …xed
for longer than one month. The degree of price rigidity is also associated with the duration as
measured by the number of months that elapse between the two consecutive nominal price changes.
In this paper, we …nd that, on average, prices take 2.7 months to change in Lesotho. These results
show that the prices in Lesotho change relatively more frequently than in the developed countries
and SA. For example, in the US, prices take 3.8 months to change and in the Euro area, 6.6 months
on average, (Gouvea, 2007) while in SA they take 4.2 months (Creamer and Rankin, 2008). But
the results are more or less in the same range with other developing countries such as Brazil and
Sierra Leone where prices in Brazil take 2.7 months to change and in Sierra Leone 3 months.
The data con…rms that there is signi…cant heterogeneity in price setting behaviour in terms of
the frequency of price change in Lesotho. This heterogeneity is realised across product categories
at both aggregated and disaggregated levels. Prices in the food sector, particularly perishable
food products, change most frequently. This could be due to the durability of perishable products
that can force retailers to change prices frequently to avoid pricing below the marginal costs.
Perishable products are also subject to seasonality. Prices in the services sector change relatively
less frequently. Services are relatively non-tradable and have lower volatility of consumer demand
hence lower frequency of price changes. In addition, most services in Lesotho such as medical and
health care, transport and education are subject to government pricing regulation and hence the
reason why their prices delay to change. One potential implication is that the monetary authority
could pay a particular attention to in‡ation in sticky sectors such as the services sector.
The results showed signi…cant time e¤ects (which include seasonal e¤ects), indicating evidence
in favour of time-dependent pricing. This kind of pricing could be caused by weather, the timing
of sales, institutional factors such as changes in regulated prices at speci…c periods of the year.
There is also signi…cant heterogeneity in price setting in terms of the size of price changes
across di¤erent sectors, which could be due to di¤erences in supply and demand conditions across
sectors. Generally price changes are small across di¤erent sectors; most price changes are less
than 5 percent. There is also evidence of asymmetric price setting behaviour across product
groups and also some degree of downward price rigidity in the price setting behaviour in Lesotho.
Asymmetries in the frequency and the size of price change suggest that the costs of price changes
cause a decrease in the size of the price change or a decrease in the frequency of the price change.
Downward price rigidity could be interpreted as a particular hesitancy of retailers to change prices.
If retailers are unwilling to reduce prices frequently, they will reduce them only when they are
facing large negative shocks. Hence why the average size of the observed price decreases will be
large.
23
Evidence also con…rms that the frequency of price changes is positively correlated with average
2 3 We
must note that this interpretation holds if the price changes are not related to end of season sales and promotions.
28
product in‡ation, suggesting that price setting behaviour is mostly driven by local in‡ation in
Lesotho. This relationship points out to state-dependent pricing rule and con…rms the predictions
by Cecchetti(1986) which states that higher in‡ation leads to more frequent price changes. Higher
product in‡ation also leads to higher frequency of price increases but lower frequency of price
decreases. Retailers in Lesotho also seem to re‡ect backward looking behaviour in terms of how
frequent they change and increased their prices. These results are in line with the predictions by
Gali and Gertler (1999). The importance of food in‡ation in price setting behaviour in Lesotho
restate the fact that food prices contribute more to the overall in‡ation in Lesotho hence also
in‡uence the frequency at which retailers change their prices. Signi…cant asymmentries in price
setting behaviour across di¤erent sectors implies that sectoral response to monetary policy is likely
to di¤er and this will a¤ect the construction of a core measure of in‡ation.
The analysis of the role of external factors from SA indicate that price setting behaviour in
Lesotho is also driven by regional conditions. Retailers in Lesotho display similar price setting
behaviour at the product level. This suggests evidence that there is market integration between
Lesotho and SA. In general, the results reveal the e¤ectiveness of SA monetary policy on pricing
conduct in Lesotho mostly through in‡ation and the frequency of price change for most products.
Therefore monetary authorities should not only take into consideration price setting behaviour in
Lesotho, but also price setting behaviour in SA in understanding monetary transmission mechanism in Lesotho.
29
10
References
Amato, J. D. and T. Laubach (2003), "Rule-of-thumb Behavior and Monetary Policy," European
Economic Review, 47,791-831.
Alvarez L. and I. Hernando (2005) “The Price setting behaviour of Spanish Firms: Evidence
from Survey Data”, European Central Bank Working Paper, no.538.
Alvarez L., Dhyne E., Hoeberichts M., Kwapil C., Le Bihan H., Lunnemann P., Martins F.,
Sabbatini R., Stahl H., Vermeulen P., And Vilmunen, J (2006) “Sticky Prices in the Euro Area:
A summary of new micro evidence”, Journal of the European Economic Association, vol. 4 (2-3):
575-584.
Álvarez, Luis J. (2008), “What Do Micro Price Data Tell Us on the Validity of the New Keynesian Phillips Curve?”Economics –The Open-Access, Open-Assessment E-Journal, vol. 2(19).
Baharad E. and B. Eden (2004) “Price Rigidity and Price Dispersion: Evidence from Micro
Data,”Review of Economic Dynamics, vol.7: 613-641.
Barro, R. J. (1972), “A theory of monopolistic price adjustment,”Review of Economic Studies,
vol. 39: 17-26.
Baudry L, Bihan H, Sevestre P and S. Tarrieu (2004) “Price rigidity in France: some evidence
from consumer price micro-data,”European Central
Bank working paper, no. 384.
Benkovskis K., L. Fadejeva and K. Kalnberzina (2011) "Price Setting Behaviour in Latvia:
Econometric Evidence from CPI Micro Data," Latvijas Banka working paper, no 1.2010.
Benkovskis K., K. Kalnberzina and L. Fadejeva (2010) "Price Setting Behaviour in Latvia:
Descriptive Evidence from CPI Micro Data," Latvijas Banka working paper, no 1/2011.
Bils M. And Klenow P. (2004) “Some Evidence on the Importance of Sticky Prices,” Journal
of Political Economy, vol. 112: 947-985.
Blinder, A.S. (1998) “Asking about Prices: A New Approach to Understanding Price Stickiness,”Russell Sage foundation: New York.
Calvo, G. (1983) "Staggered prices in a utility-maximizing framework," Journal of Monetary
Economics, vol.12: 383-398.
30
Creamer K. and N. A. Rankin (2008) “Price Setting in South Africa 2001–2007: Stylised Facts
Using Consumer Price Micro Data,”Journal of Development Perspectives, vol. 1: 93-118.
Coricelli F. and Horvath R. (2008) “Price setting and market structure: an empirical analysis
of micro data,”Institute of Economic Studies, Charles University in Prague, IES Working Paper,
no. 23.
Cecchetti, Stephen G. (1986), “The Frequency of Price Adjustment: a Study of Newsstand
Prices of Magazines,”Journal of Econometrics, vol. 31: 255-274.
Debelle G. and O. Lamont (1997) "Relative price variability and in‡ation: evidence from U.S.
cities,"Journal of Political Economy, vol. 105: 132-152.
Dhyne E. et al (2006) “Price Setting in the Euro Area and the United States: Some Facts from
Individual Consumer Price Data,”Journal of Economic Perspectives, vol. 20: 171-192.
Dias, M., D. Dias and P. Neves (2004) “Stylised Features of Price Setting Behaviour in Portugal:1992 2001,”ECB Working Paper 332.
Domberger S. (1987) “Relative price variability and in‡ation: a disaggregated analysis,”Journal of Political Economy, vol. 95: 547-66.
Fabiani, Silvia, A. Gatulli, and R. Sabbatini (2004) “The Pricing Behaviour of Italian Firms:
New Survey Evidence on Price Stickiness,”European Central Bank Working Paper, no.333.
Food and Agriculture Organisation (2007), Special Report for FAO/WFP Crop and Food Sup-
ply Assessment Missionto Lesotho. Retrieved August 17, 2011, from www.fao.org/docrep/010/ah865e/ah865e0
Fischer, S (1977) "Long-Term Contracts, Rational Expectations, and the Optimal MoneySupply Rule", Journal of Political Economy, vol. 85(1): 191-205.
Gali, J and M Gertler (1999) "In‡ation Dynamics: A structural econometric analysis",Journal
of Monetary Economics, vol. 44(2): 195-222.
Goldberg P. K Penelopi and M. M Knetter (1997) "Goods prices and exchange rates: What
have we learned?" Journal of Economic Literature, vol. 35(3): 1243-1272.
Gouvea S. (2007) “Price Rigidity in Brazil: Evidence from CPI Micro Data”, Banco Central
Do Brazil Working Paper, Series 143.
31
Greenslade J. and M. Parker (2010) "New Insights into Price Setting Behaviour in the United
Kingdom", Bank of England Working Paper, no. 395
Hall, S., M. Walsh and A. Yates (1997): “How do UK companies set prices?”, Bank of England
Working Paper, no. 67.
Klenow, P. J. and O. Kryvtsov (2008), “State-Dependent or Time-Dependent Pricing: Does
It Matter For Recent U.S. In‡ation?”Quarterly Journal of Economics, Vol. 123: 863-904.
Klenow P. J. and B.A. Malin (2010) “Microeconomic Evidence on Price setting,”NBER working paper, no. 15826
Klenow, P. J. and K. Oleksiy (2008) “State-Dependent or Time-Dependent Pricing: Does It
Matter For Recent U.S. In‡ation?”Quarterly Journal of Economics, vol. 123: 863-904.
Kovanen A. (2006) “Why do prices in Sierra Leone Change so often? A Case Study Using
Micro Level Price Data”, IMF Working Paper, no. 06/53.
Lach S. and D. Tsiddon (1992) “The behaviour of prices and in‡ation: an empirical analysis
of 17 disaggregated price data,”Journal of Political Economy, vol. 100: 349-89.
Loy, J. and Weaver, R D, (1998), "In‡ation and Relative Price Volatility in Russian Food Markets," European Review of Agricultural Economics, Oxford University Press for the Foundation
for the European Review of Agricultural Economics, vol. 25(3): 373-94.
Lucas, R., (1973), "Some International Evidence on Output-In‡ation Trade-o¤s," American
Economic Review, American Economic Association, vol. 63(3): 326-334.
Nakamura, E. and J. Steinsson (2008a) “Five Facts about Prices: A Re-evaluation of Menu
Cost Models,”Quarterly Journal of Economics, vol. 123: 1415-1464.
Parks R.W. (1978) “In‡ation and relative price variability,”Journal of Political Economy, vol.
86: 79-95.
Romer, D. (2006) "Advanced Macroeconomics" McGraw-Hill Irwin.
Sheshinski E, Weiss Y. (1977) “In‡ation and costs of price adjustment," Review of Economic
Studies, vol. 44: 287–303
32
Steinsson, J. (2003), "Optimal Monetary Policy in an Economy with In‡ation Persistence,"
Journal of Monetary Economics, 50, 1425-1456.
Taylor, J. B. (1980): "Aggregate Dynamics and Staggered Contracts", Journal of Political
Economy, vol. 88: 1-23.
Taylor, J. B. (1999) “A Historical Analysis of Monetary Policy Rules" in J. B. Taylor ed.
"Monetary Policy Rule", Chicago: University of Chicago Press, pp 319-348.
Van Hoomissen, T. (1988) “Price dispersion and in‡ation: evidence from Israel,” Journal of
Political Economy, vol. 96: 1303-14.
Vining, D.R.Jr. and T.C. Elwertowski (1976) “The relationship between relative prices and
the General Price level,”American Economic Review, vol. 66: 699-708.
Wang J., I. Masha, S. Kazuko and L. Harris (2007) "The Common Monetary Area in Southern
Africa: Shocks, Adjustment and Policy Challenges," IMF Working Paper, no. 07/158.
Wolman A.L. (2000), "The frequency and Costs of Individual Price Adjustment," Federal
Reserve Bank of Richmond Economic Quarterly, Vol. 86(4).
33
APPENDIX
item description
Wine (Non Spackling), White JC Leroux,7
50 kg Portland cement
500 single ply toilet paper per roll
Airline fare, Maseru - Johannesburg, ret
Artificial Hair (singles)
Axe
Batteries,eveready,PP10
Bostol, 100ml
Brandy,Martell VSOP, 750 ml
Charge for car maintenance, oil and grease
Chewing Gum
Cigarettes,Stuyvesant, pkt of 20
Consultation fee, adults, private doctor
Consultation fee, dentist
Consultation fee, government hospitals
Consultation with Traditional doctor
Cow milk in plastic bottle 500 ml
Dettol 125 ml
Diesel,1litre
Dried Peas
Dry cleaning of 1 pair of trousers
Dry cleaning of Ladies’skirt
Engine oil, 500 ml (diesel engine)
Engine oil, 500 ml (petrol engine)
Fabric softener, 2l
Freezer, electric
Full Cream Cow Milk in carton box 500 ml
General service of a car
High school fees
Jik 750 ml
Lagtogen,400g
Linen; 1 meter; linen 100%; plain no des
Liver, sheep,1kg
Lounge suite
Mahleu,500ml
Monthly dstv fee,allchanells
NAN,250g
NAN,400g
Novel
Oil filter for Toyota Corolla
Palmolive 125 g
Paracetamol, 500mg,
Paraffin 1 L
Paraffin 500ml
Pen
Petrol, per litre
Pipe tobacco,BB,50 g
Portable radio recorder; radio+cassette
Radio without cassette;<10W;Audio system
Records, videos,films
Refrigerator, electric, 180 litres,
Refrigirators,non-electric
Relaxer
Scotch Card
Secondary school fees
Seshoeshoe material, per meter
Spinach, per bundle
Sport bag
Steel wool 50 g
Subsc.to clubs etc.
Frequency
1,148
654
3,163
93
605
68
3,253
1,195
883
331
104
3,273
230
84
153
527
2,989
3,283
884
2,497
556
36
1,313
16
3,331
8
156
614
1,237
3,298
36
8
1,706
1,504
2,014
1,025
2,794
142
619
976
3,052
859
2,555
895
3,287
700
3,114
1,208
893
516
888
880
1,092
499
1,012
646
329
757
1,543
83
item description
drycleaning,Sportfootwear
drycleaning,suede
eggs,30s
electrical cord,0.75 mm (price per m)
eyeware,consultation fee (private clinic)
firewood,A bag
fish, fried
fish,canned,420g
flourmeal,breadflour,12.5kg
flourmeal,easybake,12.5kg
footwear repair,Half sole replacement
glassware,Drinking glass in cardboard
grilled beefsteak with potatoes,restaurant
ham,1kg
insecticides
iron stand
jelly,Moirs,50g
juice,100% fruitjuice,1litre
juice,fruitjuice,1litre
juice,fruitjuice,500ml
lactogen, 250g
lamb chops,1kg
local home brew
macaroni,fattis and monis,500g
maizemeal,chai,12.5kg
maizemeal,chai,2.5kg
maizemeal,induna,12.5kg
malt,mthombo,1kg
margarine,rama,250g
matches,One packet of ten
mayonaise,cross and blackwell,750ml
men's clothing,Jacket,Cotton 35-65%, Linen
men's footwear,Men’s casual shoes, leather
men's footwear,Sportswear(alltypes)
methylated spirit 750ml
milk,sour,500ml
mineralwater,carbonated,500ml
mineralwater,still,500ml
mint snuff,1kg
mutton chops,1kg
new car,Toyota Corolla; power steering;
new car,Toyota Yaris; power steering; co
offals, sheep,1kg
oil,sunflower,750ml
onions,1kg
oranges,1kg
other clothing, adult socks in good condition
other clothing,other second hand clothing
other fuel,Coal, one bag of 50 kg
other furniture, kitchen scheme, 3 pce
other furniture,Coal stove with 4 plates
other furniture,Gas stove with 2 burner
other furniture,Kitchen scheme,wooden
other furniture,Primus stove flame
other furniture,Wardrobe; material of fr
other household appliances,Paraffin heat
peaches,1kg
peanutbutter,blackcat,410g
plastic pipes
pork chops,1kg
34
Frequency
1,929
3,307
3,578
530
175
737
379
3,376
3,196
3,463
1,079
1,912
536
909
3,300
321
3,659
2,924
108
3,131
2,617
1,962
696
3,618
2,543
2,786
2,628
1,718
3,262
3,669
3,694
1,018
2,117
1,260
2,368
2,261
3,315
3,532
2,707
1,892
158
86
1,559
3,612
1,317
760
1,431
1,431
689
281
388
984
1,443
132
1,524
613
548
3,496
851
998
Outlet type
Hardware
Pharmacy
accommodation
alcoholic beverages
bank
bookshop
butchery
car dealer
clothing
clothing and footwear
domestic services
dry cleaning
entertainment
footwear
fruits and vegetables
fuel station
funeral parlour
furniture
furniture and hardware
grocery
hair salon
hardware
hospital
hotel
household furniture and appliance
liquor store
meals
medical services
motor parts and repairs
pharmacy
postal services
recreation and culture
restaurant
school
shoe repairs
take-away food outlet
telecommunications
transport services
Total
Maseru
1
2
2
1
0
4
7
1
2
6
1
2
1
1
2
3
0
0
0
15
1
3
0
1
4
2
0
6
6
0
1
0
2
9
1
2
1
2
92
ButhaButhe Leribe
Berea
0
0
0
2
0
1
1
0
1
0
1
1
1
3
0
0
0
1
0
2
0
0
0
1
0
0
1
1
1
1
2
1
1
0
1
0
0
0
3
7
1
1
1
0
0
0
0
0
0
5
2
2
0
0
0
0
0
3
2
0
0
0
0
0
1
0
2
2
1
1
0
0
0
0
1
1
24
36
0
0
0
0
0
2
2
0
0
5
0
1
0
0
0
1
1
0
0
5
1
3
1
0
2
2
0
0
1
1
0
0
1
2
1
0
0
1
33
Mafeteng MohalesHoek Quthing QachasNeck Mokhotlong ThabaTseka Total
0
0
0
0
0
0
1
0
0
0
0
0
0
4
0
1
0
0
0
0
4
0
0
0
0
0
0
2
0
0
0
0
0
0
1
0
0
1
1
0
0
10
3
2
1
3
1
1
24
0
0
0
0
0
0
1
1
1
0
3
2
2
12
2
1
2
0
1
5
24
0
0
2
0
0
0
3
0
1
0
0
0
0
5
0
0
0
1
0
0
2
1
1
2
0
0
0
7
0
2
0
0
0
0
6
1
2
2
2
2
1
17
0
0
0
0
0
0
2
2
1
0
0
0
0
4
0
0
3
0
0
0
3
5
6
7
4
8
7
67
1
1
2
1
2
2
13
0
0
0
2
3
1
13
0
0
0
0
0
0
1
0
0
0
0
0
0
1
0
0
1
0
2
0
14
2
2
2
3
3
1
21
1
1
0
0
0
0
2
0
1
0
1
1
0
9
1
1
1
0
1
0
14
1
0
0
1
0
0
5
0
0
0
0
0
0
1
1
1
0
0
1
0
3
0
0
0
0
0
0
4
3
2
0
0
0
1
21
1
1
0
1
1
1
9
0
0
0
0
0
0
2
0
0
0
0
0
0
1
2
1
1
3
0
0
12
28
29
27
26
28
22
345
35
Outlet type
rural
Hardware
Pharmacy
accommodation
alcoholic beverages
bank
bookshop
butchery
car dealer
clothing
clothing and footwear
domestic services
dry cleaning
entertainment
footwear
fruits and vegetables
fuel station
funeral parlour
furniture
furniture and hardware
grocery
hair salon
hardware
hospital
hotel
household furniture and appliances
liquor strore
meals
medical services
motor parts and repairs
pharmacy
postal services
recreation and culture
restaurant
school
shoe repairs
take-away food outlet
telecommunications
transport services
Total
36
urban
0
0
0
1
0
1
4
0
2
3
0
0
0
0
1
1
0
0
0
9
1
1
0
0
0
2
0
1
0
0
0
0
1
0
0
0
0
0
28
Total
1
4
4
1
1
9
20
1
10
21
3
5
2
7
5
16
2
4
3
58
12
12
1
1
14
19
2
8
14
5
1
3
3
21
9
2
1
12
317
1
4
4
2
1
10
24
1
12
24
3
5
2
7
6
17
2
4
3
67
13
13
1
1
14
21
2
9
14
5
1
3
4
21
9
2
1
12
345
Figure 3a: Monthly frequency of price changes
Figure 4a: Mothly frequency of price change
Figure 5a: Monthly frequency of price changes
Figure 3b: Frequency of price changes by product type
Figure 4b: Frequency of price change by product category
Figure 5b: Frequency of price changes by location
37
freq
Correlation coefficients between the variables
dlp
lag_dlp dlp_all_national sd_dlp
I_dlp_rural dlp_national
freq
1.000
dlp
0.078
0.000
1.000
lag_dlp
0.001
0.887
-0.211
0.000
1.000
dlp_all_national
0.060
0.000
0.162
0.000
-0.025
0.005
1.000
sd_dlp
0.014
0.122
0.389
0.000
-0.282
0.000
0.040
0.000
1.000
I_dlp_rural
0.029
0.001
0.518
0.000
-0.137
0.000
0.055
0.000
0.257
0.000
1.000
dlp_national
0.076
0.000
0.400
0.000
-0.074
0.000
0.410
0.000
0.113
0.000
0.132
0.000
1.000
Estimated regressions of frequency of price increase and product inflation
dlp
(1)
Coefficient
0.981***
(0.150)
(2)
Coefficient
1.174***
(0.116)
0.258***
(0.086)
-0.436**
(0.209)
0.003
(0.019)
(3)
Coefficient
1.143***
(0.111)
0.237***
(0.083)
-0.427**
(0.200)
-0.007
(0.019)
0.104***
(0.012)
0.399***
(0.028)
0.175***
(0.015)
0.155
12782
None
0.118***
(0.015)
0.399***
(0.024)
0.256***
(0.011)
0.265
11959
Product, District
0.067**
(0.029)
lag_dlp
I_dlp_rural
sd_dlp
dlp_national
dlp_all_national
Dnov
(4)
Coefficient
1.025***
(0.098)
0.225***
(0.079)
-0.372*
(0.191)
0.006
(0.018)
0.679***
(0.216)
3.269***
(0.400)
0.366***
(0.028)
0.201***
(0.030)
0.235***
(0.009)
0.288
11959
Product, District
(5)
Coefficient
1.034***
(0.096)
0.237***
(0.077)
-0.378**
(0.188)
-0.003
(0.018)
0.643***
(0.203)
0.120***
(0.035)
Djan
0.313***
(0.037)
Constant
0.304***
0.300***
(0.025)
(0.025)
R2
0.351
0.356
N
11959
11959
Fixed effects
Product, District,
Product, District,
Month
Month
Notes: The dependant variable is the average frequency of price increase. Regression (1) is the basic regression with only
product inflation as an independent variable. Regression (2) includes other control variables such as rural inflation, lagged
inflation and unexpected inflation (3 includes time effects. Regression (4) and (5) includes national inflation and also control for
unobserved heterogeneity. Clustered and robust standard errors are in parenthesis below the estimated coefficients. ***
Significant at 1 percent level ** significant at 5 percent level * significant at 10 percent level.
38
Frequency of price change and inflation between Lesotho and SA for durables, non-durables and services
durables
non-durables
services
Coefficient
Coefficient
Coefficient
dlp_Les
0.173
0.364
-0.092
(0.122)
(0.222)
(0.799)
dlp_Les (t-1)
-0.080
0.640
-1.182
(0.272)
(0.404)
(1.991)
dlp_SA
-0.198
0.116
-3.544
(0.191)
(0.331)
(20.914)
dlp_SA (t-1)
0.093
0.595*
-5.069
(0.108)
(0.359)
(18.720)
dlp_SA (t-2)
0.067
0.340
-2.465
(0.218)
(0.311)
(13.858)
freqsa
0.107
0.068
4.703*
(0.078)
(0.047)
(2.373)
freqsa (t-1)
0.011
0.140**
2.064
(0.061)
(0.058)
(2.797)
dlp_national
0.002
0.043
(0.670)
(0.374)
Dnov
0.500***
0.492***
0.985*
(0.098)
(0.044)
(0.517)
Djan
0.502***
0.479***
0.710*
(0.099)
(0.046)
(0.422)
Constant
0.415***
0.419***
-0.142
(0.091)
(0.037)
(0.523)
R2
0.523
0.405
0.768
N
1146
1349
124
Fixed effects
Product, month
Product, month
Product, month
Notes: The dependent variable is the average frequency of price decrease. *** Significant at 1 percent level
** Significant at 5 percent level * significant at 10 percent level. This is a within product variation regression.
Estimated regressions of frequency of price decrease and product inflation
dlp
(1)
Coefficient
-0.672***
(0.079)
(2)
Coefficient
-0.762***
(0.077)
0.061
(0.064)
0.339**
(0.140)
-0.003
(0.010)
(3)
Coefficient
-0.765***
(0.076)
0.064
(0.060)
0.368***
(0.129)
-0.010
(0.010)
0.507***
(0.026)
0.144***
(0.013)
0.111***
(0.008)
0.188
12782
None
0.506***
(0.025)
0.150***
(0.013)
0.155***
(0.008)
0.246
11959
Product, District
0.526***
(0.027)
lag_dlp
I_dlp_rural
sd_dlp
dlp_national
dlp_all_national
Dnov
(4)
Coefficient
-0.753***
(0.071)
0.060
(0.064)
0.335**
(0.139)
-0.004
(0.010)
-0.052
(0.120)
0.089
(0.333)
0.507***
(0.032)
0.149***
(0.019)
0.155***
(0.008)
0.246
11959
Product, District
(5)
Coefficient
-0.752***
(0.071)
0.064
(0.059)
0.362***
(0.128)
-0.011
(0.010)
-0.077
(0.095)
0.520***
(0.028)
Djan
0.175***
(0.018)
Constant
0.136***
0.136***
(0.012)
(0.012)
R2
0.386
0.386
N
11959
11959
Fixed effects
Product, District,
Product, District,
Month
Month
Notes: The dependant variable is the average frequency of price decrease. Regression (1) is the simplest model with only
product inflation as an independent variable. Regression (2) includes rural inflation (3) includes unexpected inflation (4)
includes time effects. Regression (5) and (6) includes national inflation and also control for unobserved heterogeneity. Standard
errors are in parenthesis below the estimated coefficients are robust and adjusted for clustering at product group level. ***
Significant at 1 percent level ** significant at 5 percent level * significant at 10 percent level.
39
Regressions of frequency of price increase and product inflation between Lesotho and SA
(1)
Coefficient
1.630***
(0.096)
(2)
(3)
perishables
Coefficient
Coefficient
Coefficient
dlp
1.665***
1.679***
1.621***
(0.104)
(0.099)
(0.144)
lag_dlp
0.233*
0.209
-0.031
(0.131)
(0.130)
(0.231)
dlp_SA
0.170
-0.192**
-0.180*
-0.604***
(0.106)
(0.094)
(0.092)
(0.183)
dlp_SA (t-1)
-0.180**
-0.115
-0.178
(0.086)
(0.084)
(0.155)
dlp_SA (t-2)
0.037
0.043
0.317***
(0.087)
(0.090)
(0.058)
freqsa_plus
0.161***
0.144***
0.002
(0.029)
(0.029)
(0.167)
freqsa_plus (t-1)
0.182***
0.173***
0.045
(0.028)
(0.028)
(0.056)
dlp_all_national
4.569***
(0.384)
Dnov
0.145***
0.423***
0.157***
0.130
(0.016)
(0.027)
(0.025)
(0.090)
Djan
0.223***
0.016
0.242***
0.161*
(0.016)
(0.025)
(0.024)
(0.090)
Constant
0.211***
0.153***
0.171***
0.206**
(0.003)
(0.004)
(0.019)
(0.087)
R2
0.240
0.288
0.350
0.419
N
6631
5997
5997
1378
Fixed effects
None
Product
Product, month
Product, month
Notes: The dependent variable is the average frequency of price increase. *** Significant at 1 percent level
** Significant at 5 percent level * significant at 10 percent level. This is a within product variation regression.
non-perishables
Coefficient
2.285***
(0.182)
0.237
(0.267)
-0.153
(0.278)
0.757***
(0.219)
0.167***
(0.050)
0.147
(0.248)
0.104**
(0.046)
0.158***
(0.044)
0.178***
(0.045)
0.230***
(0.034)
0.515
2000
Product, month
Regressions of frequency of price decrease and product inflation between Lesotho and SA
(1)
Coefficient
-1.326***
(0.070)
(2)
(3)
perishables
Coefficient
Coefficient
Coefficient
dlp
-1.442***
-1.431***
-1.581***
(0.083)
(0.080)
(0.118)
lag_dlp
0.041
-0.038
0.090
(0.123)
(0.116)
(0.206)
dlp_SA
0.010
0.089
0.025
0.116
(0.074)
(0.085)
(0.082)
(0.176)
dlp_SA (t-1)
0.072
0.062
0.201
(0.074)
(0.070)
(0.130)
dlp_SA (t-2)
0.097
0.074
-0.032
(0.077)
(0.075)
(0.082)
freqsa_minus
0.035
0.019
-0.174
(0.032)
(0.031)
(0.152)
freqsa_minus (t-1)
0.108***
0.108***
-0.020
(0.037)
(0.035)
(0.073)
dlp_all_national
0.061
(0.334)
Dnov
0.437***
0.437***
0.381***
0.346***
(0.019)
(0.027)
(0.024)
(0.116)
Djan
0.161***
0.166***
0.115***
0.196*
(0.012)
(0.021)
(0.019)
(0.113)
Constant
0.146***
0.134***
0.188***
0.190*
(0.002)
(0.004)
(0.014)
(0.111)
R2
0.295
0.314
0.421
0.487
N
6631
5997
5997
1378
Fixed effects
None
Product, month
Product, month
Product, month
Notes: The dependent variable is the average frequency of price decrease. *** Significant at 1 percent level
** Significant at 5 percent level * significant at 10 percent level. This is a within product variation regression.
40
non-perishables
Coefficient
-1.984***
(0.150)
-0.475**
(0.200)
0.275
(0.181)
-0.313*
(0.173)
0.042
(0.038)
0.242
(0.184)
0.020
(0.044)
0.424***
(0.036)
0.218***
(0.029)
0.117***
(0.019)
0.557
2000
Product, month