-
Remittances and Relative Pricesrode_645 45..61
Hiranya K. Nath and Carlos Vargas-Silva*
AbstractUsing Mexican data, this article analyzes the impact of the workers’ remittances on the cross-sectiondistribution of prices as well as on the evolution of individual relative prices over time for 272 consumeritems. The results suggest that there are important differences in the responses of relative prices to remit-tances according to various categories of these items. While the relative prices of a number of nontradableservice items such as housing consistently rise, the relative prices of several durable items such as furnituretend to fall in response to the remittance shock. Furthermore, remittances explain substantial variation inprices for a large number of consumer durables and services at various time horizons. The relative priceresponses are more volatile over time for most food items and less volatile for nonfood and service itemsreflecting different degrees of price flexibility.
1. Introduction
The economic impact of immigration on the migrant-sending countries has been thefocus of much discussion in the recent literature. One of the most well-known con-sequences of out-migration is the monetary sums sent by migrants to their familiesand friends back home. Usually the amount of the individual transfers is not large.However, the total amount of these flows can reach an enormous dimension. Currentrecorded workers’ remittance inflows into Latin America have surpassed the inflowsof foreign direct investment and official development assistance. The money thatthe migrants send home has become an important source of income for receivingfamilies.
If we treat remittances as a source of household income, other things being equal,an increase in remittances shifts the receiving household’s budget constraint outwardby the amount of the transfer and therefore has a positive impact on householdconsumption. However, the potential increase in consumption is not likely to beequal across the spectrum of all goods and services. Previous microeconomic studieshave presented mixed evidence about the uses of remittances. While some studiesargue that families receiving remittances from abroad typically spend the money ondaily needs such as food and clothing (Orozco, 2003), others argue that the remit-tance receiving households tend to spend more on investment goods like education,health, and housing, and less on food than their non-remittance receiving counter-parts (Adams and Cuecuecha, 2010; Taylor and Mora, 2006). Overall, these potentialchanges in demand for various consumption items may have an impact on the dis-tribution of relative prices.
In this article, we study the impact of remittances on the cross-section distribution ofconsumer prices in Mexico, the largest recipient of remittances in Latin America. The
* Vargas-Silva: Centre on Migration, Policy and Society, University of Oxford, 58 Banbury Road, Oxford,OX2 6QS, UK. Tel: +44-1865-274711; Fax: +44-1865-274718; Email: [email protected]: Sam Houston State University, Huntsville,TX 77341-2118, USA.Tel: +1-936-294-4760; Fax: +1-936-294-3488.
Review of Development Economics, 16(1), 45–61, 2012DOI:10.1111/j.1467-9361.2011.00645.x
© 2012 Blackwell Publishing Ltd
Author's copy
-
study of the impact of remittances on the relative prices of the receiving country isimportant because relative prices play a pivotal role in decision making by millions ofeconomic agents about consumption and production in the economy. Excessive vari-ability in relative prices leads to an inefficient allocation of resources across differentsectors of the economy and involves substantial welfare cost to the society. Thus,studying the effects of remittances on relative prices is important for understandingtheir macroeconomic consequences.
Our results also contribute to the understanding of the widely documentedpositive relation between inflation and standard deviation of relative price changes,particularly for the case of developing countries. Furthermore, information on theimpact of remittances on relative prices can be useful for the formulation of appro-priate fiscal and monetary policy in countries with large remittance inflows such asMexico.
In this study, we seek answers to the following questions about the impact of remit-tances in the receiving countries. Do remittances have any impact on the cross-sectiondistribution of prices? In particular, what are the effects of remittances on variousmoments of the price distribution? Which products or services relative prices are mostaffected by remittances? Is it possible to discern identifiable patterns in responses ofrelative prices to remittances according to various categories of consumer items such asdurable and non-durable goods, and services? To the best of our knowledge, our paperis the first study that fully analyzes the impact of remittances on the relative prices ofthe receiving country.
2. Relevant Literature and Theoretical Background
Remittances have the potential to impact a large number of variables in the recipientcountry. Therefore, the literature that tries to assess the macroeconomic impact ofremittances is varied in its scope and at times controversial in its conclusions.1 Broadly,remittances may have both beneficial as well as detrimental effects. While effects likeaugmentation of the capital stock through financing investment are conducive togrowth, other impacts such as a decrease in labor supply, may have adverse con-sequences for macroeconomic performances.
From the microeconomic perspective, some studies (especially earlier studies) con-clude that remittance transfers are spent mostly on food. For example, Orozco (2003)showed that up to three quarters of the remittance transfers are spent on food inMexico. However, some other recent studies suggest otherwise. Adams andCuecuecha (2010), using data for Guatemala, found that the remittance receivinghouseholds spend more on education and housing, and less on food than do otherhouseholds. Similarly, Taylor and Mora (2006), in a study using household level datafrom Mexico, concluded that the propensity to invest appears to be considerablylarger for households with migrants. Zarate-Hoyos (2004) showed that the Mexicanremittance receiving households devote a larger proportion of current expendituresto investment and savings and have lower income elasticities for current consumptionand for durable goods.
Do these microeconomic impacts have any implications at the macroeconomic level?At the macroeconomic level, remittances may impact a series of variables includingprices and the exchange rate. Amuedo-Dorantes and Pozo (2004), using data for 13Latin American and Caribbean countries, found that remittances cause appreciation ofthe real exchange rate. Vargas-Silva (2009) provided further evidence of real exchange
46 Hiranya K. Nath and Carlos Vargas-Silva
© 2012 Blackwell Publishing Ltd
Author's copy
-
rate appreciation for the specific case of Mexico. The impact of remittances on prices,however, has remained largely unexplored.
Our paper complements the previous literature on the macroeconomic impactof remittances in the recipient country by providing, for the first time, a detailedanalysis of the impact of remittances on relative prices. By studying the impact ofremittances on relative prices of various consumption items we not only improve ourunderstanding of the household behavior as regards to the potential uses of remit-tances but also can learn a great deal about the macroeconomic consequences ofremittances.
This line of research is also important in the context of the existing literatureon the relation between inflation and the distribution of relative price changes. Itis widely documented that there is a positive relation between inflation and thedispersion of relative price changes. However, the theoretical exposition has notbeen conclusive in regards to the causal mechanism that generates the observed rela-tionships. According to one theory propounded to explain this relationship, bothinflation and relative price variability are generated by some exogenous factor.Remittances may fit very well as that exogenous factor explaining the relationbetween inflation and various moments of relative price distribution for severaldeveloping countries.
This literature also indicates that there may be distinct patterns in responses ofthe relative prices to aggregate factors according to some identifiable commoncharacteristics. For example, Golob and Bishop (1997) reported that there are import-ant differences in the behavior of prices of durables, non-durables and services.Nath (2004) further suggested that intermediate and investment goods may responddifferently than do consumption goods to changes in aggregate factors. These studies,however, do not provide a good explanation as to why these differences exist. Aswe have argued above, remittances may have differential effects on items withgenerally identifiable characteristics such as durable goods, non-durable goods, andservices.
Some Intuition on the Effects of Remittances on Relative Prices
A recapitulation of our discussion and the findings of the existing microeconomicstudies suggest that remittances, by changing household demand for various items, willhave an impact on their prices. That the remittance income elasticity of demand variesacross consumption items is an important finding of the literature and provides theintuitive basis for the expected results of our investigation. It is, therefore, conceivablethat remittances may cause disproportionate changes in prices of various consumptionitems and that will have implications for relative prices.
If as Orozco (2003) suggested remittances are mainly spent on food, then after anincrease in remittances we may see an increase in the price of food items relative toother items. However, if remittance money is mostly spent on services such as educa-tion, health, and housing then we may observe an increase in the price of these servicesrelative to other items. This should be especially the case if, as Adams and Cuecuecha(2010) argued, remittance receiving households also decrease their spending on foodcompared to other households. Similarly, Zarate-Hoyos (2004) argued that remittancereceiving households have lower income elasticities for consumption of durable goods.If this holds true, then we can expect remittances shocks to decrease the relative priceof durable goods.
REMITTANCES AND RELATIVE PRICES 47
© 2012 Blackwell Publishing Ltd
Author's copy
-
This intuition about changes in relative prices is further reinforced by the fact thatremittance transfers are often targeted towards a specific household consumption. Forexample, some studies (e.g. Amuedo-Dorantes et al., 2007) suggested that remittancesare targeted towards healthcare expenditures. This type of targeted transfers willchange the consumption patterns of the remittance receiving households and, in aggre-gate, may have implications for relative prices.
The above discussion is based on the implicit assumption that prices are fully flex-ible, but as it is a well known, most prices are not fully flexible. Thus, even ifremittances increase demand for various consumption items proportionately, theremay be disproportionate changes in prices with implications for relative prices.Hence, although it is intuitively clear that remittances will cause changes in relativeprices, it is not clear a priori how they will affect relative prices of individual items.Therefore, we now turn to an empirical investigation of the effects of remittances onprices in Mexico.
3. Methodology and Data
Methodology
As it is standard in most macroeconomic studies of this type, we use a vector auto-regressive (VAR) model. We estimate a seven variable system that includes industrialproduction (IP), remittance inflows (REM), the overall consumer price index (CPI)(P), CPI of individual consumption item (PICI), the interest rate (I), M2 measureof money (M), and the exchange rate (X). All variables, but the interest rate, areused in natural logarithms.2 Since the data are of monthly frequency, we include12 lags of each variable as well as a constant in each equation of the VAR. In orderto compute impulse response functions and variance decompositions, we use conven-tional Cholesky decomposition to obtain orthogonal residuals. We use the followingordering of the variables in the model: IP, REM, P, PICI, I, M, and X. In this case,we assume that output is not contemporaneously affected by shocks to the othervariables, while the exchange rate is contemporaneously affected by shocks to allthe other variables. We further assume that shocks to remittances have contempor-aneous effects on individual prices, while shocks to prices affect remittances with alag.3
We estimate the VAR model recursively including one individual consumer itemprice at a time. Thus, for each of the 272 individual prices, we estimate an equationrelating that price series with the macroeconomic variables in the VAR.That is, for eachprice index we estimate an equation of the following form:
p A B L p D L Xit i i it i t it= + ( ) + ( ) + ε , (1)
where pit is the log of price index of good i, Ai is the constant term, Xt is the vector ofmacroeconomic variables, and Bi and Di are lag polynomials.
It is very important to recognize that because we include the general price level(measured by overall CPI) in the VAR model, the impulse responses represent neteffects of the remittance shock on individual prices. That is, these responses are to beinterpreted as effects of remittances on individual prices over and above their effects onthe general price level. Thus, the price of an individual item can be thought of as a“relative price”, relative to the overall price level.
This methodology has been used by several other studies on relative prices. Forexample, Balke and Wynne (2007) and Lastrapes (2006) used VARs to estimate the
48 Hiranya K. Nath and Carlos Vargas-Silva
© 2012 Blackwell Publishing Ltd
Author's copy
-
impact of shocks to variables like money supply, interest rates and productivity onrelative prices using data for the USA. Both papers include commodity prices andoverall CPI in the VARs in order to see the impact of these shocks on relative prices.As with the current study, Balke and Wynne (2007) used short-run restrictions andinclude the commodity prices one at a time, while Lastrapes (2006) preferred long-runrestrictions to identify the shocks.
The inclusion of standard macro variables such as output, money supply, interest rate,and exchange rate is primarily motivated by a desire to control for the effects of thesevariables on individual markets through well known channels.As we mentioned above,aggregate factors such as output, money supply and interest rates have been shown onprevious studies to have an important impact on relative prices (Lastrapes, 2006; Nath,2004). Moreover, the exchange rate with its possible impact on the prices of tradableand nontradable goods also has important implications for relative prices (Betts andKehoe, 2008).
In this structure of the model, for each individual price series we assume no feedbackto or from other individual price series.4 We estimate impulse responses of all individualprices for one through 24 months to a shock to remittances and generate the cross-section distributions at various time horizons after the shock. We also calculate variousmoments of these cross-section distributions at different time horizons. Finally, weestimate variance decompositions of individual prices to identify items for which vari-ations in prices are explained by remittances.
Data
We use monthly CPI data for 272 consumption items for the period 1996:01–2007:06.Alist of these individual price series is included in the Appendix (Table A1). The baseyear for the price series is June 2002. Seasonal adjustments have been made to theprices using the Census X-11 method. The choice of the sample period is dictated bytwo considerations. First, although data on remittances are available before 1996, theyare more reliable only for recent years. Second, we want to avoid the years of theMexican financial crisis, particularly 1994 and its immediate aftermath.
In addition to disaggregate prices, we also obtain data on overall CPI; industrialproduction that represents output; monetary aggregate, M2; short-term interest rate;and the nominal exchange rate that is defined as Mexican pesos per US dollar. Notethat although we would like to include gross domestic product as a measure of output,the data are not available at the monthly frequency. The interest rate is the Mexicangovernment’s three-month bond rate (CETES interest rate). Finally, we use totalfamily remittances as a measure of Mexico’s inward remittances. All the data areobtained from Mexico’s Central Bank.
A first glance at the data Figure 1 plots the smoothed cross-section distribution of thesample averages (i.e. averaged over the sample period) of the 272 price change seriesalong with a theoretical normal distribution with the same mean and variance as thecross-section distribution of price changes. The price changes are the first log differ-ences of seasonally adjusted price series.The sample average price change for each itemis positive with an unweighted mean of 0.009225 for the cross-section distribution of272 price changes. Note that although a mean of 0.009225 looks small, if one convertsit into percentage annual rate, it equals 11.07% (= 0.009225 ¥ 12 ¥ 100). A comparisonwith the normal distribution reveals that the actual distribution is positively skewedand leptokurtic.
REMITTANCES AND RELATIVE PRICES 49
© 2012 Blackwell Publishing Ltd
Author's copy
-
To shed light on the nature of the relation between remittances and prices, wecalculate correlation coefficients between remittance growth and price changes foreach of 272 items over the sample period. Figure 2 plots the cross-section distributionof the correlation coefficients, along with a theoretical normal distribution with thesame mean and variance as the empirical distribution. The distribution has a positivemean and is slightly negatively skewed indicating that a relatively larger number ofprice changes have positive correlation with remittance growth.
In Figure 3 we plot various moments of the cross-section distribution of pricechanges along with remittance growth for the entire sample period. The first,the second, and the third order moments have positive relation with remittancegrowth while kurtosis has negative correlation. The skewness and remittancegrowth appear to have the strongest correlation. Although these figures capture theevolution of the cross-section distribution of price changes over the sample period, the
0
50
100
150
200
250
300
-.004 .000 .004 .008 .012 .016 .020
ActualNormal
Den
sity
Figure 1. Cross-section Distribution of Sample Averages of Individual Price Changes
0
2
4
6
8
10
-.3 -.2 -.1 .0 .1 .2 .3
ActualNormal
Den
sity
Figure 2. Cross-section Distribution of Average Correlations Between Price Changesand Remittance Growth
50 Hiranya K. Nath and Carlos Vargas-Silva
© 2012 Blackwell Publishing Ltd
Author's copy
-
-.4-.2.0.2.4
-.02
.00
.02
.04
.06
.08
9798
9900
0102
0304
0506
07
rem
itta
nce
grow
thav
erag
e pr
ice
chan
geremittance growth
average price change
Mon
ths:
199
5:02
- 20
07:0
6C
orr
= 0
.039
-.4-.2.0.2.4
.01
.02
.03
.04
.05
.06
.07
95
96
95
9695
96
95
9697
9899
0001
0203
0405
0607
rem
itta
nce
grow
thst
d de
v. o
f pri
ce c
hang
es
remittance growth
std dev. of price changes
Mon
ths:
199
5:02
- 20
07:0
6C
orr
= 0
.002
-.4-.2.0.2.4
-15
-10
-50510
9798
9900
0102
0304
0506
07
rem
itta
nce
grow
thsk
ewne
ss o
f pri
ce c
hang
es
remittance growth
skewness of price changes
Mon
ths:
199
5:02
- 20
07:0
6C
orr
= 0
.091
-.4-.2.0.2.4
04080120
160
9798
9900
0102
0304
0506
07
rem
itta
nce
grow
thku
rtos
is o
f pri
ce c
hang
es
remittance growth
kurtosis of price changes
Mon
ths:
199
5:02
- 20
07:0
6C
orr
= -0
.039
Fig
ure
3.T
ime
Seri
esof
Sele
cted
Mom
ents
ofth
eC
ross
-sec
tion
Dis
trib
utio
nof
Pri
ceC
hang
esan
dR
emitt
ance
Gro
wth
REMITTANCES AND RELATIVE PRICES 51
© 2012 Blackwell Publishing Ltd
Author's copy
-
contemporaneous correlations between remittance growth and the respectivemoments reported at the bottom of each graph do not reveal much about the dynamicrelation between remittances and cross-section distribution of price changes.Therefore,we examine the dynamic relation between remittances and relative prices, controllingfor a standard set of macroeconomic variables.
4. Results
Remittances and the Distribution of Relative Price Responses
In this subsection and the next, we present the results based on impulse responses ofrelative prices to a one standard deviation shock to remittances, generated from theestimation of the VAR model discussed above. Remember that these responses are tobe interpreted as effects of remittances on individual prices over and above their effectson the general price level. This interpretation has two important implications. First, anegative response does not necessarily imply that the price decreases in response to aremittance shock. It may simply be that the increase in that particular price is lowerthan the overall price rise. Second, the response may be small in magnitude but to takefull stock of the absolute price response we need to combine the responses of thegeneral price level and the individual price. The following results are important to theextent that they change relative prices which, as we have discussed above, are crucialfor consumption and production decisions of the millions of economic agents in theeconomy.
Figure 4 plots the fractions of negative and positive relative price responses to aremittance shock at various horizons. Until the ninth month, the proportion of negativerelative price responses is not only higher than that of positive relative price responsesbut also the difference rises. After the horizon of 9 months, however, these two pro-portions converge. Past the 21st month, the proportion of positive relative priceresponses exceeds the proportion of negative relative price responses. It seems that inthe short run, innovations to remittances have a negative impact on a larger number of
.2
.3
.4
.5
.6
.7
.8
2 4 6 8 10 12 14 16 18 20 22 24
negativepositive
months
frac
tion
s
Figure 4. Fractions of Negative and Positive Price Responses at Various Horizons
52 Hiranya K. Nath and Carlos Vargas-Silva
© 2012 Blackwell Publishing Ltd
Author's copy
-
relative prices. A remittance shock has positive effects on a larger number of relativeprices only in the longer horizon.
Figure 5 provides yet another look at the evolution of the distribution of relativeprice responses to a remittance shock. Figure 5 plots the mean, and the 5 and 95percentile of the cross-section distribution of price responses over the horizons from 1to 24 months. As we can see, the mean response becomes positive after eighteenmonths. Note that the lower 5 percentile decreases steadily until the eighth month,increases sharply between months 9 and 13 and increases very slowly after the 14thmonth. The 95 percentile increases until the 14th month and then becomes somewhatvolatile.
In Figure 6, we present the cross-section distribution of relative price responses athorizons of 1, 6, 12, and 24 months, along with normal distributions with the same meansand variances as the empirical distributions. Although the mean of contemporaneousprice responses (i.e. at a horizon of one month) to a remittance shock is positive it isinfinitesimally small. Then the mean responses are negative until the time horizon ofnineteen months when it turned positive. For example, the annualized percentage mean(unweighted) price response at a horizon of one month is 0.03. At an horizon of 6months, it is -1.06%, at 12 months it is -0.59%, and at a horizon of 24 months it is0.29%.
The dispersion of price responses seems to increase from horizon one to horizon six.The distribution seems to be positively skewed at a horizon of 1 month while it seemsto have a longer tail to the left at a horizon of 6 months. At horizons of 12 and 24months, the distribution is almost symmetric with some fluctuations in between.Although the distribution is leptokurtic at all horizons, the kurtosis decreases overtime.These changes in the distribution of price responses indicate that relative prices ofvarious goods and services change at different horizons as a result of changes inremittances.
In Figure 7, we plot higher order moments along with the mean of impulseresponses for horizons of 1–24 months. As discussed above, there is a substantialliterature that indicates that positive relations between mean and higher order
-.008
-.006
-.004
-.002
.000
.002
.004
.006
.008
.010
2 4 6 8 10 12 14 16 18 20 22 24
mean responses5 percentile95 percentile
months
resp
onse
s
Figure 5. Mean, 5 Percentile and 95 Percentile of Price Responses to a Remittance Shockat Various Horizons
REMITTANCES AND RELATIVE PRICES 53
© 2012 Blackwell Publishing Ltd
Author's copy
-
moments are robust empirical results. Although remittances do not seem to generatean empirically plausible relation between mean and standard deviation of relativeprice changes, they may have significant explanatory power for the relation betweenmean and skewness. In this case (mean and skewness), the estimated correlationcoefficient is 0.60.
Remittances and Individual Relative Price Responses
While the previous discussion provides useful details about the impact of remittanceson the cross-section distribution of relative prices, it is also important to look at theresponse of individual relative prices to the remittance shock over time. For spaceconsiderations we do not include the responses of all 272 consumer prices to theremittance shock. However, a few words regarding the responses of representativegoods and services prices are in order.
Two food items, beef steak and fresh pasteurized milk, with relatively higherweights among food items in the consumption basket (1.14 and 1.86 respectively)differ in their responses to the remittance shock.5 For beef steak, a delicacy, as aresult of the remittance shock, demand seems to increase steadily raising relativeprice over time. The price of fresh pasteurized milk, a necessity, meanders aroundzero indicating that the price change is similar to the change in the general pricelevel. As we will see below, a large percentage of variation in relative prices of
0
100
200
300
400
500
600
700
-.03 -.02 -.01 .00 .01 .02 .03
Den
sity
horizon = 1
0
50
100
150
200
250
300
-.04 -.02 .00 .02 .04
Den
sity
horizon = 6
0
50
100
150
200
250
-.03 -.02 -.01 .00 .01 .02 .03
Den
sity
horizon = 12
0
50
100
150
200
250
-.03 -.02 -.01 .00 .01 .02 .03
actualnormal
Den
sity
horizon =24
Figure 6. Cross-section Cumulative Distribution of Price Responses to a RemittanceShock
54 Hiranya K. Nath and Carlos Vargas-Silva
© 2012 Blackwell Publishing Ltd
Author's copy
-
-.001
6
-.001
2
-.000
8
-.000
4
.000
0
.000
4
.002
5
.003
0
.003
5
.004
0
.004
5
24
68
1012
1416
1820
2224
mea
nst
d. d
ev
corr
=-0
.21
mean
standard deviation
-.001
6
-.001
2
-.000
8
-.000
4
.000
0
.000
4
-8-4048
24
68
1012
1416
1820
2224
mea
nsk
ewne
ss
corr
=0.
60
mean
skewness
-.001
6
-.001
2
-.000
8
-.000
4
.000
0
.000
402040608010
0
24
68
1012
1416
1820
2224
mea
nku
rtos
is
corr
=-0
.30
meankurtosis
Fig
ure
7.M
ean,
Stan
dard
Dev
iatio
n,Sk
ewne
ss,a
ndK
urto
sis
ofP
rice
Res
pons
esto
aR
emitt
ance
Shoc
k
REMITTANCES AND RELATIVE PRICES 55
© 2012 Blackwell Publishing Ltd
Author's copy
-
durable items like living room furniture, modular equipment, and glazed pottery isexplained by remittances. The relative prices decrease in response to the remittanceshock for all these three items. For consumption items that provide nontradable ser-vices such as housing, house for rent, electricity, cars, and restaurants and have largerweights in the consumption basket, relative prices rise steadily over time in responseto the remittance shock.
The impulse responses of relative prices for two additional service items, educationand health that have been shown to be affected by remittances in the previous micro-economic studies (Adams, 2005; Taylor and Mora, 2006) provide interesting results.Therelative price of education (all levels) responds very little and meanders around zero.The relative price of doctor visits increases steadily in the first 5 months and thendecreases until it starts rising again after 18 months.
In order to get an idea how remittances affect the volatility of prices of individualconsumption items in Mexico, we order the items by the variability of impulseresponses over horizons of 1–24 months (measured by standard deviation over thisperiod). We report the standard deviation of the top 20 items in Panel A of Table 1 andthe standard deviation of the bottom 20 items in Panel B of the same table. We alsoreport the mean responses and the number of times the responses to a remittanceshock are positive. Note that the items which are most volatile in their price responsesalmost exclusively include food items such as vegetables and fruits.
In contrast, the items that are least volatile in their price responses include a com-bination of nontradable services and tradable goods. For example, education at variouslevels shows very little variability in price responses. The average changes in prices arealso relatively small for this group. Among others, visits to cafeterias, nightclubs, res-taurants, and bars exhibit low volatility in their responses to a remittance shock. Theseresults are not surprising. The prices of food items are generally more flexible thanthose of nonfood and service items and they respond very quickly to changes in themarket conditions. The prices of education, restaurant food, etc., are slow to adjust tothose changes and therefore exhibit less volatility.
Variance Decompositions
We further resort to variance decomposition to calculate the percentage of variationin individual prices that is explained by remittances. We order the consumption itemsaccording to the percentage of price variation explained by remittances from largestto the smallest at horizons of 6, 12, and 24 months and report the top 20 items inTable 2. Remittances seem to explain relatively larger proportion of price variationsfor a number of durable items at various horizons. For example, at a horizon of 6months, significant variations in prices of stove, glazed pottery, antecomedores (spe-cialty furniture), radio and recorders, and other domestic electronic gadgets seem tobe explained by remittances. Further, at horizons of 12 and 24 months, modularequipment, cassettes, CDs and living room furniture are some additional durableitems which are included among the top 20 items for which significant price vari-ations are explained by remittances. Substantial variation in prices of nontradableservices such as car insurance, phone line, other entertainment, and houses for rent isalso explained by remittances at various time horizons. Note that there are few fooditems in Table 2.
To summarize, a positive shock to remittances lowers relative prices for a largernumber of consumption items in the short run. At longer horizons, however, a largernumber of relative prices rise in response to the remittance shock. The relative prices
56 Hiranya K. Nath and Carlos Vargas-Silva
© 2012 Blackwell Publishing Ltd
Author's copy
-
Tabl
e1.
Item
sO
rder
edby
Var
iabi
lity
inIm
puls
eR
espo
nses
over
1–24
Mon
thH
oriz
onsa
Pan
elA
:Top
20ite
ms
Pan
elB
:Bot
tom
20ite
ms
Item
sSt
anda
rdde
viat
ion
Mea
n(%
)P
ositi
vere
spon
ses
Item
sSt
anda
rdde
viat
ion
Mea
n(%
)P
ositi
vere
spon
ses
(1)
(2)
(3)
(4)
(1)
(2)
(3)
(4)
1Se
rran
och
ili0.
0183
–4.1
011
253
Shoe
sm
ales
0.00
03–0
.54
42
Cha
yote
0.01
358.
7717
254
Caf
eter
ias
0.00
030.
3019
3G
reen
tom
atoe
s0.
0132
–6.9
29
255
Und
erw
ear
girl
s0.
0003
–0.3
16
4O
nion
s0.
0124
–23.
620
256
Can
dy0.
0003
0.69
245
Lem
ons
0.01
23–5
.22
925
7O
ther
shoe
s0.
0003
–0.3
85
6To
mat
oes
0.01
18–1
.34
825
8E
duca
tion
,sho
rtca
reer
0.00
030.
1519
7P
inea
pple
s0.
0103
–0.7
39
259
BB
Q,c
ooke
d0.
0003
0.65
248
Avo
cado
es0.
0103
3.33
1126
0E
duca
tion
,col
lege
0.00
03–0
.33
39
Pum
pkin
s0.
0089
–0.3
112
261
Jack
ets
and
coat
s0.
0003
–0.9
40
10C
hick
peas
0.00
8610
.80
2026
2N
ight
club
0.00
03–0
.63
111
Ora
nges
0.00
76–1
0.31
426
3C
otto
npa
nts
mal
es0.
0003
–0.1
98
12Po
tato
es0.
0073
3.61
1626
4R
esta
uran
ts0.
0003
–0.0
112
13O
ther
chili
s0.
0073
–5.1
69
265
Und
erw
ear
mal
es0.
0002
–0.6
21
14C
arro
ts0.
0071
2.35
1326
6Te
xtbo
oks
0.00
020.
1917
15P
apay
as0.
0071
1.06
1226
7E
duca
tion
,pri
mar
y0.
0002
–0.4
61
16B
anan
as0.
0065
2.30
1326
8T
ires
0.00
020.
0813
17C
hile
pobl
ano
0.00
64–2
.39
926
9B
ars
0.00
020.
0816
18W
ater
mel
ons
0.00
591.
7913
270
Edu
cati
on,i
nter
med
iate
0.00
020.
2721
19C
ucum
ber
0.00
581.
5414
271
Edu
cati
on,k
inde
rgar
ten
0.00
02–0
.13
920
Subw
ay0.
0055
3.93
1327
2B
eaut
ysa
lon
visi
t0.
0002
–0.0
86
Not
e:a C
olum
n(2
)re
port
sth
est
anda
rdde
viat
ion
ofth
eim
puls
ere
spon
ses,
colu
mn
(3)
repo
rts
the
mea
nof
the
impu
lse
resp
onse
s(a
nnua
lized
perc
entr
ate)
,and
colu
mn
(4)
repo
rts
the
num
ber
ofpo
siti
vere
spon
ses.
REMITTANCES AND RELATIVE PRICES 57
© 2012 Blackwell Publishing Ltd
Author's copy
-
Tabl
e2.
Item
sO
rder
edby
Per
cent
age
ofV
aria
tion
Exp
lain
edby
Rem
ittan
cesa
Hor
izon
=6
Hor
izon
=12
Hor
izon
=24
Item
sV
aria
tion
(%)
Item
sV
aria
tion
(%)
Item
sV
aria
tion
(%)
(1)
(2)
(3)
(4)
(5)
(6)
1H
ouse
deod
oran
t49
.09
Oth
erliq
uors
62.2
6Te
quila
66.9
42
Oth
erliq
uors
41.4
1Po
wde
rm
ilk55
.98
Mod
ular
equi
pmen
t56
.26
3Te
quila
39.7
2H
ouse
deod
oran
t55
.27
Can
ned
soda
s54
.36
4To
wel
s39
.50
Toile
tpa
per
52.9
1O
ther
elec
t.do
mes
tics
50.5
35
Car
insu
ranc
e38
.08
Tequ
ila52
.31
Saus
age
47.1
96
Pow
der
milk
36.2
7M
odul
areq
uipm
ent
51.2
4W
ine
45.8
47
Win
e34
.09
Win
e44
.13
Shir
tsfo
rm
ales
45.5
18
Stov
e33
.66
Pho
nelin
e43
.28
Liv
ing
room
furn
itur
e45
.24
9P
hone
line
32.5
9A
ntec
omed
ores
40.2
9C
asse
ttes
and
CD
s44
.95
10Fo
reig
nbu
ses
32.1
4D
ress
esfo
rfe
mal
es40
.26
Toile
tpa
per
43.7
911
Hat
s31
.82
Gla
zed
pott
ery
39.4
5O
ther
liquo
rs43
.78
12G
laze
dpo
tter
y31
.62
Shir
tsfo
rm
ales
38.2
9O
ther
spic
es43
.59
13T-
shir
tsfo
rba
bies
31.0
0To
wel
s37
.55
New
spap
ers
42.5
014
Ant
ecom
edor
es29
.35
Stov
e36
.16
Oth
eren
tert
ainm
ent
42.4
815
Toile
tpa
per
29.0
4C
anne
dso
das
35.7
7N
apki
ns41
.29
16R
adio
and
reco
rder
s27
.65
Cas
sett
es&
CD
s35
.59
Gal
zed
pott
ery
39.9
917
Oth
erpe
rson
alhy
gien
e27
.23
Nap
kins
34.7
3D
ress
esfo
rfe
mal
es39
.46
18O
ther
serv
ices
27.1
2N
ewsp
aper
s32
.30
Subw
ay38
.69
19B
eer
26.5
1T-
shir
tsfo
rba
bies
31.7
3Po
wde
rm
ilk37
.63
20O
ther
elec
t.do
mes
tic
26.1
7Sh
irts
and
T.fo
rbo
ys31
.58
Hou
ses
for
rent
37.1
8
Not
e:a C
olum
ns(2
),(4
)an
d(6
)re
port
the
perc
enta
geof
the
vari
atio
nin
pric
eof
the
prod
uct
expl
aine
dby
rem
itta
nces
.
58 Hiranya K. Nath and Carlos Vargas-Silva
© 2012 Blackwell Publishing Ltd
Author's copy
-
that remittances seem to affect positively include mostly those items that providenontradable services such as house for rent, housing, electricity, cars, and restaurants.These positive effects are more prominent at longer horizons. The relative prices ofdurable items, in contrast, tend to respond negatively to the remittance shock. Theimpulse responses of most food prices are highly variable over time. Furthermore,remittances explain substantial variation in the prices of consumer durables such asantecomedores, living room furniture, glazed pottery, domestic electronic gadgets,modular equipment, radio and recorders; and services such as car insurance, phone line,entertainment, subway, and houses for rent at various horizons after a shock.
6. Concluding Remarks
Previous microeconomic studies have suggested that workers’ remittances may haveimportant effects on the spending patterns of the receiving households. Hence, at theaggregate level remittance flows may affect prices of different goods and services in away that has implications for relative prices. In order to examine this possibility, weconsider a fairly general VAR model of the Mexican economy, in which we include, inaddition to standard macro variables and remittances, consumer price indices of 272items.
Our results indicate that in the short run a positive shock to remittances lowersrelative prices for a larger number of consumption items. At longer horizons,however, a larger number of consumer prices rise in response to the remittanceshock. More importantly, our results suggest important differences in the responses ofrelative prices according to various categories of consumer items. While there is aconsistent rise in relative prices for a number consumption items providing nontrad-able services such as housing, electricity, cars, and restaurants, the relative prices ofseveral durable items such as furniture, glazed pottery, and modular equipment tendto respond negatively to the remittance shock. Furthermore, remittances explain sub-stantial variation in prices for a large number of consumer durables and services atvarious time horizons. The relative price responses are more volatile over time formost food items and less volatile for nonfood and service items reflecting differentdegrees of price flexibility. These results are consistent with the previous microeco-nomic studies that suggest that remittance transfers are spent on a wide range ofgoods and not on food alone.
Overall, our results clearly indicate that remittances affect prices of various con-sumption items, having important implications for relative prices. It will be far-fetchedto draw any definitive conclusions on their implications for the overall price level andinflation. However, that relative prices change as a result of remittances is in itself animportant result because changes in relative prices affect the decisions of consumersand producers, and therefore have implications for the allocation of resources and theoverall well-being. To avoid distortions owing to these changes in relative prices, thepolicymakers may take into account the differential effects of remittances on prices ofvarious categories of consumer items and accommodate fiscal and monetary policiesaccordingly.
REMITTANCES AND RELATIVE PRICES 59
© 2012 Blackwell Publishing Ltd
Author's copy
-
Appendix
Table A1. List of Consumer Prices Series
Corn tortillas Candy Doctor visit Oil (cooking) Local callingCorn meal Gelatin Surgery Apples Long dist. nat. callCorn Other food. Dental care Bananas Long dist. int. callSweet bread Carnitas. Haircut Oranges Phone lineWhite bread Roasted chicken. Beauty salon visit Avocadoes Domestic serviceSandwich bread BBQ, cooked. Hair products Mangos Kitchen furnitureCakes, pies Beer Perfumes and lotions Papayas AntecomedoresNoodles Tequila Hand soap Lemons StovePopular cookies Other liquors Toothpaste Grapes Water heatherOther cookies Rum Deodorant Melons Living room furnit.Wheat flour Brandy Skin lotion Watermelons Dinning tableCereal Wine Shaving machine Pears SofaRice Cigarettes Other hygiene Peaches Bed furniturePoultry Shirts males Toilet paper Grapefruits FridgeChicken (pieces) Underwear males Diapers Pineapples Washing machineWhole chicken Socks males Pads (females) Guavas Other electro.Pork pulp Pants males Napkins Tomatoes FansPork chops and fat Suits males Facial tissue Potatoes IronPork loin Other pants males Colectivo Onions BlenderPork legs Clothe males Urban bus Other legumes Modular equipmentBeefsteak Shirts females Taxi Green tomatoes Radios/recordersGround beef Underwear fem. Subway Pumpkins Light bulbsCow remnants Pantyhose Foreign bus Serrano chili MatchesSpecial beef cuts Cotton pants fem. Air transportation Carrots BatteriesLiver Other pants fem. Cars Other chilis CandlesOther cow Suits females Bicycles Chile poblano BroomsHam Dresses females Lubricant oil Lettuce and cabbage Glazed potterySausage Skirts females Tires Chickpeas CookwareChorizo Cotton pants boys Other car accessories Prickly pears Utensils (plastic)Other stuffed food Other pants boys Car batteries Chayote Other utensilsDry meat Shirts boys Car insurance Cucumber BedspreadBacon Dresses girls Other car spending Green beans Other home textilesOther fish Underwear boys Car maintenance Beans Bed sheetShrimp Underwear girls Tolls Dry chili BlanketsMojarra Socks boys Parking Other dry legumes TowelsOther seafood Dresses babies Private edu., college Canned juices CurtainsBass and grouper Shirts babies Private edu., primary Processed chili Laundry detergentHuachinango Jackets and coats Private edu., high Canned vegetables Soap for dishesTuna Hats Private edu., int. Tomato paste/soup House deodorantCanned seafood Sweaters child Private edu., short Other canned fruits PesticidesFresh milk School unif. Boys Private edu., kinder Fruits for babies AntibioticsPowder milk School unif. Girls Textbooks Sugar AnalgesicsEvaporated milk Tennis shoes Other books Instant coffee Cardiovascular md.Fresh cheese Shoes females Notebooks Toasted coffee Nutritional supp.Yogurt Shoes males Pens, pencils Canned sodas Birth control pillsOaxaca cheese Shoes boys Hotels Mayonnaise/mustard Gastrointestinal Md.Sour cream Other shoes Cinema Chicken/salt spices ExpectorantChihuahua cheese Baggage Nightclub Other seasoning Other medicinesOther cheeses Watches, jewelry Other entertainment French fries Flu medicineIce cream Houses for Rent Sport clubs Concentrated drinks First aidYellow cheese Housing Sport events Chocolate Glasses (eye)Butter Electricity Newspapers Toys Cameras and relatedEggs Domestic gas Magazines Cassettes and CDs Musical instrumentsSport accessories Restaurants Cafeterias Memorial services Other servicesLunch places Bars
60 Hiranya K. Nath and Carlos Vargas-Silva
© 2012 Blackwell Publishing Ltd
Author's copy
-
References
Adams, Richard H., Jr and Alfredo Cuecuecha, “Remittances, Household Expenditure andInvestment in Guatemala,” World Development 38 (2010):1626–41.
Amuedo-Dorantes, Catalina and Susan Pozo, “Workers’ Remittances and the Real ExchangeRate: A Paradox of Gifts,” World Development 32 (2004):1407–17.
Amuedo-Dorantes, Catalina, Susan Pozo, and Tania Sainz, “Remittances and Healthcare Ex-penditure Patterns of Populations in Origin Communities: Evidence from Mexico,” Integration& Trade Journal 27 (2007):159–184.
Balke, Nathan and Mark Wynne. “The Relative Price Effects of Monetary Shocks,” Journal ofMacroeconomics 29 (2007):19–36.
Barth, Marvin and Valerie Ramey, “The Cost Channel of Monetary Transmission,” NationalBureau of Economic Research working paper 7675 (2000).
Betts, Caroline and Timothy Kehoe, “Real Exchange Rate Movements and the Relative Priceof Non-traded Goods,” National Bureau of Economic Research working paper 14437 (2008).
Davis, Steven and John Haltiwanger, “Sectoral Job Creation and Destruction Responses to OilPrices Changes,” Journal of Monetary Economics 48 (2001):465–512.
Golob, John and David Bishop, “Inflation and Relative Price Variability: Durables vs.Nondurables and Services,” Federal Reserve Bank of Kansas City working paper 12 (1997).
Lastrapes, William, “Inflation and the Distribution of Relative Prices: The Role of Productivityand Money supply Shocks,” Journal of Money, Credit and Banking 38 (2006):2159–98.
Nath, Hiranya, “Relative Importance of Sectoral and Aggregate Sources of Price Changes,”Applied Economics 36 (2004):1781–96.
Orozco, Manuel, “Remittances, the Rural Sector, and Policy Options in Latin America,” USAIDand BASIS Collaborative Research Support Program (CRSP) (2003).
Ruiz, Isabel and Carlos Vargas-Silva, “To Send, or Not to Send: That is the Question. A Reviewof the Literature on Workers’ Remittances,” Journal of Business Strategies 26 (2009):65–90.
Sims, Christopher, “Macroeconomics and Reality,” Econometrica 48 (1980):1–49Sims, Christopher, James Stock, and Mark W. Watson,“Inference in Linear Time Series Models
with some Unit Roots,” Econometrica 58 (1990):113–44.Taylor, J. Edward and Jorge Mora, “Does Migration Reshape Expenditures in Rural House-
holds? Evidence from Mexico,” World Bank Policy Research working paper 3842 (2006).Vargas-Silva, Carlos, “The Tale of Three Amigos: Remittances, Exchange Rates and Money
Demand in Mexico,” Review of Development Economics 13 (2009):1–14.Zarate-Hoyos, German, “Consumption and Remittances in Migrant Households: Toward a
Productive use of Remittances,” Contemporary Economic Policy 22 (2004):555–65.
Notes
1. For a survey, see Ruiz and Vargas-Silva (2009).2. Following Balke and Wynne (2007), we include these variables in levels. Sims(1980) andSims, Stock and Watson (1990) are among the first to suggest that even if the variables areunit root processes, they should be included in levels. They argue that the goal of a VARanalysis is to determine the interrelationships among variables, not to determine the parameterestimates.3. This paper focuses on the demand side effect of remittances on relative prices. However, if aportion of the remittances is used for investment they may also affect the supply side. Byincluding industrial production in our VAR model, we indirectly consider the supply side of themarket. But introducing a direct channel through which remittances may affect supply is outsidethe scope of the current study.4. Balke and Wynne (2007), Barth and Ramey (2000) and Davis and Haltiwanger (2001) makea similar assumption about feedback in their papers.5. The weights are percentage consumption shares in the market basket of a typical urbanhousehold in Mexico.
REMITTANCES AND RELATIVE PRICES 61
© 2012 Blackwell Publishing Ltd
Author's copy