Tradable and non-tradable inflation in consumer prices in Pakistan 1

Abstract

According to monetarists, inflation is always and everywhere a monetary phenomenon. It may, however, have supply side drivers; including global commodity prices. In order to have better understanding of inflation dynamics in a small open economy like Pakistan, we need some new sub-indices within CPI basket. In this paper, we have proposed and estimated new subgroup indices within CPI basket of Pakistan for monthly data from July 1991 to June 2018. These sub-indices include those for ‘services’ and ‘goods’ prices. Within goods group, we have further bifurcated the sub-indices for ‘tradable’ and ‘non-tradable’ goods prices. Tradable goods price index is further bifurcated into ‘food and energy’ and ‘non-food and non-energy’ tradable sub groups. On average, goods account for two thirds of our CPI basket and remaining one third belongs to services. Non-tradable goods and services comprises of three-fourths of overall consumer basket, while one-fourth are tradable goods. Share of tradable goods in CPI basket has increased over time. Inflation rate in tradable goods prices is volatile and lacks persistence whereas services prices are found relatively stable and highly persistent in Pakistan. Inflation in tradable goods basket is found to lead inflation in non-tradable prices.

Keywords

Index numbers price level inflation tradable and nontradable goods

1. Introduction

Pakistan Bureau of Statistics (PBS) collects retail prices of a ‘periodically fixed’ set of goods and services from urban parts of the country on monthly basis and compiles consumer price index (CPI) for the country.2 PBS classifies different items into various groups.3 In order to better understand inflation dynamics, in a small open economy like Pakistan, there is a need of more subgroups within CPI basket. Inflation is not necessarily always and everywhere a monetary phenomenon. Hanif et al. [1] observed that global commodity prices also matter in explaining inflation in small open economy such as Pakistan. And that, trade openness has played a significant role in dampening inflation in Pakistan economy [2]. Moreover, contribution of supply side drivers to inflation can be ‘changing’ over time as has been observed in the case of Pakistan by Hanif [3]. Furthermore, there can be different inflation drivers in different subgroups as Hanif and Iqbal [4] established that monetary policy does not matter for ‘administered prices’ in Pakistan. Two main subgroups – tradables and nontradables – deserve attention to better understand sources of inflation, and effectiveness of monetary and exchange rate management thereon, in Pakistan. By tradables we mean goods which are imported and/or exported at prices usually determined in the markets away from Pakistan like meat. By nontradable we generally mean a) goods that are seldom traded internationally like fresh milk, and b) services like haircut.

We know that consumer prices for tradables items depend upon international price of underlying commodity and importing country’s exchange rate (in addition to trade levies in trading partners). We cannot directly observe this by simply analyzing overall price index (like CPI) or looking at the available sub-indices of CPI of Pakistan. A separate ‘tradable goods’ prices index may be useful for analytical purposes to get the idea of first round impact of exchange rate movements on tradeable goods’ price changes in a country. Even within ‘tradable goods prices, the speed of pass through of exchange rate to food and energy prices may be different compared to for non-food non energy tradable goods prices. Services prices sub index may be relatively less sensitive to exchange rate movements compared to tradable goods’ prices. As mentioned earlier, PBS does not provide these types of sub-indices for prices and thus we cannot carry out such analysis. This study is an attempt to fill this gap.4

We divide the overall CPI basket into separate lists of aforementioned sub groups as shown in the conceptual framework section. All the items in the CPI basket are divided into goods and services. Goods are further bifurcated into tradable and non-tradable goods. Following Dwyer [6], any good (in the CPI basket) for which Pakistan (internationally) trades at least 10 percent of domestic production (of the same product) is classified as tradable one. Within tradable goods’ list, we separate the food and energy items and non-food-non-energy items. Non-traded goods list is merged with the services list to have a price index of ‘non-tradable goods and services’. The new sub-indices are constructed using individual commodities’ weights in the respective CPI basket. For each of the subgroups the weights are normalized to 100 so that each index will have 100 for the base year (2007–08) index number.

We find that the new sub groups, we propose in this study, have different weights in the different CPI baskets. On average, goods account for two thirds of baskets and remaining one third belongs to services. Non-tradable goods and services basket comprises of three-fourths of overall consumer basket while one-fourth are tradable goods. Share of traded goods in typical Pakistani consumer’s consumption increased over the time from one-fifth in early 1990s to one-fourth recently. Tradable goods basket is actually dominated by food and energy related items, as the share of non-food non-energy items in it is merely 6.6 percent.

Aligned with our expectations, month on month (MoM) inflation in tradable goods prices is found to be the most volatile (as was observed by Knight and Johnson [7] for the economy of Australia) and services items price changes are found to be the most stable in Pakistan. MoM inflation in tradable goods prices does not show any persistence against all the new subgroups exhibiting inflation inertia. Service prices inflation is highly persistent in Pakistan. We observe causality from tradable prices year on year (YoY) inflation to non-tradable prices YoY inflation in Pakistan. We observe higher inflation in non-tradable sector, over the period of study, compared to tradable one. In view of relatively higher inflation in non-tradable sector, over the period of study, one can say that Pakistan is an ‘internally competitive’ economy.

Organization of the remaining part of the paper is as follows. In Section 2, we review studies pertaining to construction of consumer price sub-indices for tradable and non-tradable sectors. In Section 3, research gaps in the subject area are discussed as motivation. In Section 4, conceptual framework used to disaggregate the CPI basket of Pakistan in different subgroups is described. In Section 5, we present methodology applied and the data used to estimate the proposed sub-indices for different subgroups. We discuss some observations from the estimated sub-indices in Section 6. Some caveats of the study are listed in Section 7 along with a few suggestions on future research direction. Last section is reserved for summary and conclusion.

2. Literature review

Usually all the national statistical agencies in the world disseminate consumer price indices of subgroups similar to those published by PBS. These sub-indices are source of important insights about inflation rate dynamics at sectoral level. Relative prices in different subsectors guide resource allocation amongst different sectors. In the absence of any sub-indices for tradable and non-tradable sectors, investors and policy makers are left to assess internal and external competitiveness situation either on the basis of exchange rate alone and/or with the help of import and export unit value indices. The former may give partial picture (i.e. excluding the global prices’ dynamics) while the latter is only a type of wholesale/purchase price index.

Relative prices of tradable and non-tradable goods determine the allocation of resources between the foreign and home markets that have implications for net exports and thus external competitiveness of the country. Misallocation of resources between internal and external sectors can result in widening trade deficit. Despite this level of importance of tradable and nontradables prices data for understanding and analyzing sectoral prices’ dynamics, not all the countries’ national statistical agencies provide sub-indices of tradable and nontradables prices. This has resulted in a few research studies on the tradable and non-tradable prices dichotomy. In the following we critically review the methodologies used by these studies and see which one is the most preferred and why.

Salter [8] defined traded goods as exports and imports replacements. In case one follows this approach to classify goods as traded, then exports industries are described according to their exports orientation and imports industries are described according to their import substitutability. The degree of exports orientation and imports substitutability requires judgement. There is an added difficulty in this approach pertaining to availability of prices of selected ‘export oriented’ and ‘import substitution’ industries.

Goldstein and Officer [9] elaborated tradable/non-tradable criterion for 12 industrial countries including Austria, France, Italy, Sweden, Ireland, Japan, and Netherland. For this purpose, they derived the data from National Accounts of OECD countries and Yearbook of National Accounts Statistics by United Nations for the period of 1950–73. They picked actual commodities’ trade along with the potential tradables to bifurcate commodities/industries. Due to selection of potential tradables, this approach runs the risk of over selection of goods as tradables.

Dwyer [6] studied the tradable and non-tradable dichotomy by analyzing the Australian data to understand the allocation of resources between tradable and non-tradable sectors based on their relative prices. She used the Australian input output data of 108 industries for the period of 1974 to 1987. She developed a system for the classification of goods among tradables and non-tradables. She classified goods having 10% or more of its production internationally traded as tradables.

De Gregorio et al. [10] investigated causes of disinflation of 1970s and 1980s in 14 OECD counties. They analyzed inflation in tradable and non-tradable commodities separately by utilizing sectoral data. They took the data from IMF for the period between 1970 and 1985 for 14 OECD countries. They bifurcated tradable and non-tradable sectors by setting a criterion: if ratio of total exports of a sector to its total output is above 10% then that sector would be treated as tradable. They employed this method for four sectors i.e. manufacturing, mining, agriculture and services. Their results indicated manufacturing, mining and agriculture sector as tradable and services as non-tradable.

Knight and Johnson [7] worked on classification of tradable and non-tradable sector for Australia, initially by using a system developed by Dwyer [6]. They used the data for the period between 1977 and 1995. In their study they observed changes in composition and size of tradables (by disaggregating into importable and exportable components). They worked on identification of individual industries within tradable and non-tradable sectors of Australian economy. Price indices for tradables and non-tradables revealed volatility of prices in tradables.

Dixon et al. [11] explored the tradable and non-tradeable inflation in consumer prices index for New Zealand. In defining the tradeable and non-tradeable goods, they used the threshold provided in the study of Dwyer [6] and created tradeable and non-tradeable indices. They also conducted stability test of tradeable and non-tradeable commodities using different thresholds i.e. 10, 15 and 20 percent by looking at the addition and subtraction of various industries into/from the tradeable or non-tradeable sectors. The result of the study indicated that only a few industries were changing their status from tradable to non-tradable and vice versa by changing the threshold.

Amador and Soares [12] differentiated tradable and non-tradable sectors to analyze indicators for competition in the Portuguese economy for the period between 2000 and 2009. For commodities differentiation for tradability/non tradability, they adopted the criteria of exports to sales ratio and set the benchmark of 15%. This approach may be difficult to use for a developing country (like Pakistan) as it is easy to obtain output data (from national income accounts) compared to aggregated sale of every product in CPI.

Attewell and Crossan [13] studied the tradeable sector for New Zealand. In their study they defined tradeable and non-tradeable goods with reference to production measure of GDP. They defined tradeable goods as the goods facing competition in foreign market, while the non-tradeable are goods which are not facing any foreign competition. For doing so, they set a benchmark: goods are tradeable if 10 percent or more of (any that goods’ industry) output is exported or at least 20 percent of (any industry’s) inputs are imported. In order to set the benchmark they took input-output table data for March 2007. Following this methodology, they categorized industries fulfilling at least one of the both criteria’s (either export oriented or import competitive) as tradeable. The analysis was done on 106 industries. This was a direct method for defining tradeable goods. On the other hand, the indirect method has higher thresholds: industries and their subsidiaries exporting more than 25 percent are considered as tradables. By applying both criteria they classified industries to be tradable/non-tradable.

Canas and Gouveia [14] applied a new methodology i.e. FiPEI (Final product Exports and Imports) criterion for allocation of commodities to be tradable or non-tradable for Portugal. They took the data from Banco de Portugal and Instituto Nacional de Estatística for the period between 2010 and 2013. According to this criterion, imports were added to exports to compile trade volume of a specific industry and then trade to output ratio (TOR) was calculated. If TOR was greater than 10 percent then goods of the specific industry were taken as tradable.

Johnson [15] worked on tradable and non-tradable goods and services indices to analyze their inflation separately for US. He took the data from Bureau of Labor Statistics and Bureau of Economic Analysis for the period between 2010 and 2015. According to him the goods sold in location other than the place of production are tradable and vice versa are non-tradables. Accordingly, services, in his view, are non-tradable (except a few) in almost all over the world. For empirical analysis, he opted for the methodology used by Statistics New Zealand i.e. 15 percent as threshold for a good to be defined as a tradable one. However, he tried different percentage levels as threshold from 1 to 15 percent. He found 11 percent to be the best threshold. He observed that 1 percent shift in threshold in either way results in shifting 13 industries from one status (tradable/non-tradable) to another. He also advocated that the goods are tradable, if their prices are same in different location (as arbitrage makes the prices same). He found that tradables are characterized by lower inflation rate in US as compared to non-tradables in the examined period.

3. Motivation

In Pakistan, there is no tradable/non-tradable bifurcation of CPI basket. It is essential to dichotomize the items in CPI basket as tradable/non-tradable to estimate tradable and non-tradable items’ inflation rates separately. This study is an attempt to fill this gap. Relative behaviour of prices for tradables and non-tradables will serve as guide for resource allocation in the country between the two sectors. This will also help in analyzing country’s international trade competitiveness through ratio of non-tradable price to internationally traded good’s prices. Moreover, this will also facilitate in estimating the difference between exchange rate pass through to inflation for tradable and nontradable sectors, in addition to exploring the determinants of tradables and nontradables inflation in the country.

4. Conceptual framework

We start with PBS list of goods and services in the CPI basket. We divide the list of items in CPI basket into goods and services groups. Based on the findings of De Gregorio et al. [10] and Johnson [15] services are considered as nontraded all over the world. We also consider services as non-tradables barring a few items classified as services in CPI basket.

However, goods are bifurcated into non-tradable goods (i.e. goods produced locally) and tradable goods. Tradable items are further divided into a) food and energy and b) non-food-non-energy subgroups. Services are clubbed with domestically produced goods to form another group named as ‘non-tradable goods and services’. See Fig. 1.

Figure 1.

Conceptual framework.

5. Data and methodology

PBS selects goods and services in the CPI basket and classifies them in different groups within CPI basket based on international practices: shares of different goods and services in the household expenditures observed in periodically conducted family budget surveys are used as weights in the CPI basket. As pattern of expenditures on different goods and services change over time, PBS reviews and revises its list of commodities as well as associated weighs periodically. Last CPI basket change was made in 2007–08. Based on the survey results, 487 goods and services were included in CPI basket with respective expenditure shares for each item. PBS uses base year prices to construct different indices. Prior to this PBS was using 2000–01 base year prices for CPI basket of 374 items for that period. Before 2000–01 basket, PBS was using 1990–91 as base year with basket of 462 items. We use last three CPI baskets of goods and services for this study.5

As already discussed in the conceptual framework, we divide the list of items in CPI basket into goods and services. Not all the items in CPI basket are straightforward to label as either good or service. For example, electricity and gas are provided as service but is also a part of industrial goods. We classify electricity and gas as goods. So these types of items are categorized carefully by getting insights from PBS classification in National Income Accounts. In the next step, goods are further distributed into two subgroups – tradable and non-tradable. This classification was a major challenge due to data paucity at goods level. Accordingly, following Dwyer [6] – and some other authors such as De Gregorio et al. [10], Knight and Johnson [7] and Dixon et al. [11] – any item for which sum of imports and exports is greater than or equal to 10 percent of its output in Pakistan is treated as a tradable.

Three inference-based criteria are applied in goods for which trade to output ratio (TOR) could not be estimated (due to non-availability of output data). First: those goods for which we could not find any data on export/import after exhausting all known sources are assumed to be non-tradable.6 Second; some goods are classified as non-tradable after confirming from their producers that there is no international trade7 of these goods. Lastly: goods for which domestic production is either nil, or very close to nil, are considered to be imported and thus tradable.8 All the goods which do not qualify at least one of criteria set above are classified purely based on our own intuition; and the proportion of such goods turned out to be of 19.60% weight of overall CPI basket. Further, tradable goods are divided into two sub groups i.e. non food non energy and food and energy goods on the basis of their existing status in PBS classification (in respective subgroups).

Once goods are classified into different subgroups, new (sub) indices are made using Laspeyre’s formula as given below:

$I_{n}=\frac{\sum(P_{in}/P_{i0})\times W_{i0}}{\sum W_{i0}}\times 100$

Where, $I_{n}=$ Index for the $n^{\rm th}$ period, $P_{in}=$ price of $i^{\rm th}$ item in the $n^{\rm th}$ period, $P_{i0}=$ price of $i^{\rm th}$ item in the base period, $W_{i0}=$ weight of the $i^{\rm th}$ item in the base period.

We normalized each subgroup’s weight to be 100 to generate an index. With the help of these sub-indices, it became straightforward to calculate sub group wise MoM and YoY inflation rates.

We crosschecked our sub-indices by matching the weighted average of the relevant sub-indices with the ‘CPI General’ as estimated by PBS. For the list of new sub-indices estimated in this study, see Table 1.

Table 1
List of new sub-indices estimated for Pakistan in this study

New sub-groups	Description
GCPI	Good consumer price index
SCPI	Services consumer price index
TCPI	Tradable consumer price index
NTCPI	Non-tradable consumer price index
TCPI-NFNE	Tradable consumer price index – nonfood non energy
TCPI-FE	Tradable consumer price index – food & energy

We estimated some basic statistics like mean, standard deviation etc. for inflation pertaining to all these new sub-indices. We also estimated a couple of advance statistics like inflation persistence and inflation diffusion index. Former is calculated to understand how different are underlying data generating processes of different subgroups indices of CPI while the latter helps assessing the spread of inflation/deflation across the commodities within the basket of respective subgroup of the CPI basket [16, 17]. These basic and advanced statistics highlight the importance of making these sub-indices for meaningful economic analysis which otherwise is impossible.

In order to estimate the degree of inflation persistence we use sum of autoregressive coefficients (SARC) from an estimated AR(p) model following Hanif et al. [16]. In case SARC for an underlying inflation series is found positive and statistically significant, inflation is said to be persistent. Inflation diffusion index (IDI) is based on month-on-month (MoM) change in the prices of the commodities in the respective subgroup and is calculated as follows:

No. of items with postive change in price – No. of items with negative change in price/No. of items for which prices are reported for the $\displaystyle\quad\textit{given month}\ast 100.$

Thus, it is the difference between the share of items with increasing prices (i.e. depicting inflation) and the share of items with decreasing prices (i.e. depicting deflation) in the given basket amongst the commodities for which prices have been reported. We do not observe the prices of certain commodities (like the seasonal items) during specific months every year. To calculate inflation diffusion index we consider the commodities for which prices have been reported (irrespective of changed or not) during the month under review.

6. Data

We have used data on prices, international trade and output of the country to estimate tradable and nontradable inflation rates for Pakistan economy. Monthly commodity wise prices data of last three CPI baskets (1990–91, 2000–01 and 2007–08) is used in this study. PBS assigns monthly weights, based on the findings of respective household expenditure surveys, to each commodity to calculate monthly CPI index. These monthly weights are not available on PBS website. We used (annual) available weights of each commodity, which we assume are average of 12 months weights. Base year commodity prices are also publically unavailable. We used commodity wise average of 12 month prices of the first year of each basket as base year price to estimate new sub-indices. For classification of goods into tradable and nontradable, we have to consider another data set – commodity wise imports and exports in Pakistan. For this purpose we used commodity wise merchandise imports/exports (at HS-8 level disaggregation) extracted from PBS. Specifically, we have used annual average imports and exports for the period covered in different CPI baskets. For commodity wise output of items in CPI goods list we have utilized different sources for agriculture and industry related output. Crop wise estimates are obtained from ‘Agriculture Statistics of Pakistan’ – a publication of Ministry of National Food Security and Research – for the relevant years. Detailed dataset on manufactured items was used as obtained from PBS.

7. Results and discussion

Estimated sub-indices of newly proposed sub groups – Goods CPI (GCPI), services CPI (SCPI), non-tradables CPI (NTCPI), tradables CPI (TCPI), non food non energy tradables CPI (TCPI_NFNE), and food and energy tradables CPI (TCPI_FE) – are presented in the Appendix of this study (Table 6). These indices are presented with same base (2007–08) despite the fact that we had different base baskets (1990–91, 2000–01, and 2007–08) to work out these sub-indices. We used splicing method to convert overall sub-indices to same base sub-indices. Table 2 shows basket wise number (percent share) of items in each of the sub-indices we have worked out in this study.

Table 2
Number (percent share) of items in new sub-groups

CPI baskets	GCPI	SCPI	NTCPI	TCPI	TCPI_NFNE	TCPI_FE	Overall
1990–91	403 (87.2)	59 (12.8)	345 (74.7)	117 (25.3)	68 (14.7)	49 (10.6)	462 (100)
2000–01	319 (85.3)	55 (14.7)	238 (63.6)	136 (36.4)	73 (19.2)	63 (16.8)	374 (100)
2007–08	397 (81.5)	90 (18.5)	318 (65.3)	169 (34.7)	101 (20.7)	68 (13.9)	487 (100)

Table 3

Basket wise and average weights of new sub-indices

CPI baskets	GCPI	SCPI	NTCPI	TCPI	TCPI_NFNE	TCPI_FE
1990–91	73.48	26.52	78.61	21.39	5.28	14.13
2000–01	65.75	34.25	76.60	23.40	5.81	17.59
2007–08	63.55	36.45	74.47	25.53	8.65	16.88
Average	67.59	32.41	76.56	23.44	6.58	16.20

Figure 2.

Tradable CPI and average unit value indices of exports and imports.

Our tradable basket weights are different during different periods depending upon the list of goods we categorized as tradable during the different periods of time and breadth of the CPI basket used by PBS during different regimes (Table 3). We have also considered variation in list of tradable goods over the period of time by using three different baskets. Number and Share of traded goods increased from 1990–91 basket to 2007–08 basket.

7.1 Verification of new sub-indices

We crosschecked our sub-indices by matching the weighted average of the relevant sub-indices with the ‘CPI General’ as estimated by PBS. The difference between these two indices was small: $-$ 1.40 to 0.83 percent.9 In case of YoY inflation, such difference ranged from $-$ 0.75 to $+$ 0.57 percentage points.10

There is another way to assess the validity of our estimate of the tradeable prices index that is through comparison of unit value indices of exports and imports for Pakistan. In order to test the validity of our TCPI, we compared it with the trade-weighted average of unit value indices of exports and imports (AUXM) for Pakistan. Figure 2 shows that both the TCPI and AUXM co-move. Resultantly, both the TCPI and AUXM11 based YoY inflation also co-move, as shown in Fig. 3.

Table 4
Some basic and advanced statistics on MoM change in sub-indices within CPI

New subgroup $\to$	GCPI	SCPI	NTCPI	TCPI	TCPI_NFNE	TCPI_FE
Mean	0.68	0.61	0.66	0.67	0.51	0.72
Standard deviation	1.25	0.43	0.87	2.06	1.07	2.57
Jarque bera	30.14	109.35	48.90	4.11*	429.28	3.66
Breaks	Jul-13	Mar-97	Oct-13	Oct-13	Apr-97	Nov-13
Sum of autoregressive coefficients	0.51*	0.80*	0.66*	$-$ 0.01	0.41*	$-$ 0.38
IDI_Average	0.35	0.34	0.34	0.31	0.29	0.29
IDI_Minimum	$-$ 0.04	$-$ 0.08	$-$ 0.01	$-$ 0.32	$-$ 0.26	$-$ 0.43
IDI_Maximum	0.63	0.70	0.63	0.72	0.87	0.87
${}^{*}$ : Positive and significant at 5%.
a: Some Basic Statistics on YoY Change in Sub-indices within CPI
New subgroup $\to$	GCPI	SCPI	NTCPI	TCPI	TCPI_NFNE	TCPI_FE
Mean	8.49	7.52	7.60	6.67	6.40	9.09
Standard deviation	5.36	2.62	3.97	6.02	5.43	7.56
*: Significant at 5%.
b: Granger non-causality analysis
Null hypothesis			Obs.	Lag length	F-stat	Prob.
IDI_NTCPI does not Granger Cause NTCPI (YoY inflation)			298	13	2.18	0.01*
NTCPI (YoY inflation) does not Granger Cause IDI_NTCPI			298	13	0.54	0.90
IDI_TCPI does not Granger Cause TCPI (YoY inflation)			298	13	1.60	0.08**
TCPI (YoY inflation) does not Granger Cause IDI_TCPI			298	13	1.29	0.22
${}^{}$ : Significant at 5 percent. ${}^{*}$ : Significant at 10 percent.

Figure 3.

Tradable Inflation (YoY, in %) and YoY change in Trade Weighted Average Unit Value Indices of Export and Import (in %).

One may ask why we need tradable price index when we already have UVIX and UVIM. It is clear from the figure that, at times, the proxy of tradable goods prices based upon UVINX and UVIM is erratic as it is based on a) whole sale type prices and b) it includes prices which are not necessarily part of CPI (like defense imports and exports of sports goods). In addition to this, for the items which are common in CPI and in UVIX and UVIM, prices of goods traded internationally are inherently closer to those tradables prices which are in Wholesale Price Index (WPI) basket rather than to those in CPI basket.

7.2 Analyzing new sub-indices

In the following, we have provided some basic statistics on new estimated sub-indices of CPI basket of Pakistan. In addition to the basic statistics, we have also estimated some advanced statistics which are necessary to get some understanding about the underlying ‘data generating process’ of the inflation rates from these sub-indices and to assess the spread of inflation/deflation in respective subgroup basket. These advanced statistics include the estimates on ‘inflation persistence’ and ‘inflation diffusion index’ for each of these new sub-indices.

Figure 4.

Non-tradable vs Trdable Goods Inflation (YoY, in %).

Figure 5.

Goods CPI , Services CPI vs Overall CPI Inflation (YoY, in %).

All the results from basic and advance statistics indicate intuitive behavior of the newly constructed sub-indices within the overall CPI basket of Pakistan (Table 4) as discussed below.

Tradable goods inflation rate is found to be most volatile in the country and its distribution is close to normal indicating almost equal probability for traded goods’ prices increase and decrease. The same is also evident from the minimum and maximum values of IDI for tradable goods as compared to those for nontraded goods.

In case of nontradable it is only one month in 27 years (or 324 observations) that IDI showed negative values meaning the number of goods/services for which prices decreased were greater than the number of goods/services for which prices increased. Such behaviour was relatively more common in case of traded goods, particularly the traded food and energy goods. A simple analysis (Granger non-causality test) suggests that IDI is a leading indicator for YoY inflation, for respective baskets, for the case of Pakistan [17]. In this study, we found (using Granger non-causality test) that IDI is a leading indicator for YoY inflation both for tradable and nontradable baskets for Pakistan (Table 4b).

Moreover, tradable goods inflation does not show any persistence. On the contrary, service prices’ inflation is observed to be relatively stable and it exhibits persistence like in case of overall inflation in Pakistan [16].

Table 4a exhibits that YoY inflation in tradable sector has been lower than in non-tradable sector – like in USA[15]. However, this inflation in tradable sector is found to be the most volatile. This can be validated from the recent most basket. Tradable and non-tradable inflation (YoY) is plotted in Fig. 2 that depicts inflation rate in tradables is more volatile than non-tradables. The standard deviation for the non-tradable inflation is 3.5 compared to 5.7 for the tradable inflation rate for the recent basket. These are not much different compared to those for overall period of this study as shown in Table 4a.

Table 5

Granger non-causality analysis

Null hypothesis	Observation	Lag length	F-Stat	Prob.
$\Delta$ TCPI does not Granger Cause $\Delta$ NTCPI	295	13	2.29	0.01
$\Delta$ NTCPI does not Granger Cause $\Delta$ TCPI	295	13	1.29	0.21

Figure 6.

Exchange rate YoY changes vs tradeable/nontradable inflation ratio.

Another observation is that after July 2013 inflation rate in the prices of tradable goods remained significantly lower than that in non-tradables goods. The former has even fallen below zero in FY15 and FY18 (Fig. 4). We can attribute this volatility in the prices of tradeable goods mainly to fluctuations in exchange rate of Pakistan (Fig. 7a and b) as argued by Engel [18] – in addition to higher volatility ‘nature’ of global/traded commodity prices as observed by Hanif [3].

Like in USA, tradables in Pakistan are characterized by lower inflation rate compared to non-tradables over the period of study. There can be different reasons for higher inflation rate in the non-tradable goods market compared to those which are traded internationally. From Choudhary et al. [19] we know that manufacturing goods producing firms in Pakistan are operating in a monopolistically competitive environment. We know that firms trading internationally are facing relatively more competition than firms operating locally only. From here one can conjecture why inflation rate is higher in non-tradable goods related CPI basket. As should be, inflation rate in tradables is found to be commoving relatively strongly with changes in exchange rate compared to that of inflation rate in non-tradables, particularly after FY13 (Fig. 5). Our findings also support the observation of Jacob and William (2014) that exchange rate is one of the most important determinant of tradable inflation.

One can also observe that inflation rate for services sector is lower than for the goods sector. Even more interesting observation is (if we plot a time series graph of goods and services sector): the services sector inflation has been immune from the shocks in goods sector inflation emanating from the tradable goods prices’ volatility.

Prior to 2013, when PKR depreciated, tradable inflation was higher than non-tradable (Fig. 4). The story changed after 2013 when exchange rate appreciated and remained stable (barring some depreciation in 2015). Inflation in tradables followed it. We also plotted the ratio of tradable to non-tradable items inflation rate with exchange rate changes (Fig. 6). It shows that after 2013, inflation rate in tradables has been less than inflation rate in non-tradables in Pakistan (ratio being less than unity). Following PKR depreciation in 2015, ratio increased again above unity and reached as high as 1.60. After this, Pakistan had stable exchange rate in until December 2017. We would like to interpret it as Human Resources induced Dutch Disease (HRIDD). Pakistan has not been having this Dutch Disease from any ‘physical resource abundance’ rather it was induced (from exchange rate stability resulting) from significant amount of workers’ remittances. Workers’ remittances financed above four-fifths of country’s (goods and services) trade deficit during FY13-FY16, for example.

Figure 7.

a: Exchange rate and tradeable prices volatality; b: Exchange rate changes and tradeable inflation volatality.

We find that tradable prices inflation rate Granger causes inflation in non-tradable CPI sector of the country,12 and not the other way round (Table 5).

Table 6

New sub-indices for Pakistan from July 1991 to June 2018 (2007–08 $=$ 100)

Time period	GCPI	SCPI	TCPI	NTCPI	TCPI_NFNE	TCPI_FE
Average weights	67.59	32.41	23.44	76.56	6.58	16.20
Jul-91	23.76	28.55	23.06	25.75	36.13	19.95
Aug-91	24.08	28.82	23.41	26.05	36.23	20.32
Sep-91	24.08	29.07	23.61	26.06	36.30	20.54
Oct-91	24.36	29.30	23.78	26.36	36.42	20.72
Nov-91	24.26	29.55	23.68	26.37	36.43	20.60
Dec-91	24.10	29.77	23.67	26.28	36.69	20.54
Jan-92	24.34	29.98	23.96	26.49	36.87	20.85
Feb-92	24.52	30.20	24.19	26.68	37.31	21.03
Mar-92	24.81	30.41	24.77	26.86	37.55	21.66
Apr-92	25.46	30.62	24.97	27.52	37.61	21.87
May-92	25.94	30.83	24.51	28.21	37.86	21.30
Jun-92	25.96	31.06	24.79	28.21	37.93	21.60
Jul-92	26.33	31.25	25.34	28.47	38.13	22.20
Aug-92	27.00	31.46	26.71	28.79	38.41	23.73
Sep-92	26.86	31.66	26.50	28.78	38.37	23.50
Oct-92	26.95	31.87	26.32	29.00	38.48	23.26
Nov-92	27.04	32.11	26.53	29.10	38.60	23.49
Dec-92	26.97	32.34	27.10	28.92	38.76	24.11
Jan-93	26.93	32.56	26.71	29.06	38.87	23.64
Feb-93	27.08	32.89	26.92	29.26	39.17	23.83
Mar-93	26.95	33.12	26.54	29.31	39.37	23.36
Apr-93	27.67	33.34	26.81	30.02	39.58	23.62
May-93	28.54	33.55	27.18	30.86	39.74	24.02
Jun-93	28.82	33.73	28.61	30.76	40.10	25.60
Jul-93	29.05	33.99	28.54	31.09	41.65	25.24
Aug-93	29.64	34.20	28.99	31.61	42.00	25.69
Sep-93	29.63	34.64	29.16	31.68	42.02	25.89
Oct-93	30.20	35.03	30.20	32.06	44.25	26.67
Nov-93	30.32	35.25	30.66	32.11	45.02	27.07
Dec-93	29.96	35.51	30.48	31.88	45.82	26.72
Jan-94	30.29	35.75	31.30	32.03	46.61	27.52
Feb-94	30.88	35.98	32.22	32.42	46.86	28.53
Mar-94	30.83	36.23	31.78	32.58	46.88	28.02
Apr-94	31.91	36.52	31.80	33.76	47.23	27.97
May-94	31.98	36.75	30.76	34.22	47.80	26.67
Jun-94	32.11	37.07	31.64	34.18	48.04	27.64
Jul-94	32.80	37.43	32.82	34.63	48.23	28.96
Aug-94	33.22	37.71	33.52	34.91	48.31	29.76
Sep-94	33.60	38.00	33.84	35.29	48.48	30.09
Oct-94	34.40	38.29	35.30	35.74	48.36	31.79
Nov-94	34.77	38.53	35.94	35.99	48.61	32.48
Dec-94	34.94	38.80	36.20	36.17	48.75	32.75
Jan-95	35.76	39.09	38.11	36.49	48.38	35.02
Feb-95	35.52	39.42	37.29	36.60	48.50	34.05
Mar-95	35.76	39.71	37.53	36.87	48.62	34.31
Apr-95	35.65	40.01	35.26	37.55	48.86	31.65
May-95	36.20	40.31	35.06	38.26	49.28	31.35
Jun-95	36.29	40.58	35.85	38.19	50.91	31.96
Jul-95	36.91	40.88	36.26	38.78	51.64	32.30
Aug-95	37.95	41.15	37.99	39.38	52.37	34.16
Sep-95	38.05	41.52	38.25	39.52	53.44	34.26
Oct-95	38.01	41.80	37.46	39.81	52.01	33.62
Nov-95	38.21	42.09	37.43	40.11	52.61	33.47
Dec-95	38.50	42.32	38.93	40.01	53.57	35.02
Jan-96	38.25	43.00	38.32	40.16	53.87	34.27
Feb-96	38.50	43.31	38.43	40.47	54.04	34.36
Mar-96	39.07	43.63	39.04	40.96	54.56	34.96

Table 6, continued
Time eriod	GCPI	SCPI	TCPI	NTCPI	TCPI_NFNE	TCPI_FE
Apr-96	39.74	44.01	39.49	41.62	55.14	35.37
May-96	40.06	44.42	38.26	42.44	55.55	33.89
Jun-96	39.82	44.75	38.56	42.21	55.73	34.20
Jul-96	40.60	45.12	39.34	42.88	56.01	35.04
Aug-96	41.23	45.47	40.06	43.40	57.84	35.54
Sep-96	41.71	45.79	41.95	43.40	60.67	37.20
Oct-96	42.31	46.22	41.95	44.14	59.03	37.50
Nov-96	42.56	46.78	42.59	44.37	60.01	38.05
Dec-96	42.76	47.31	43.45	44.47	60.98	38.87
Jan-97	43.67	47.96	43.94	45.45	61.76	39.29
Feb-97	44.28	48.29	45.12	45.79	61.91	40.62
Mar-97	43.74	48.69	44.05	45.70	64.83	38.85
Apr-97	45.24	49.12	44.20	47.31	62.51	39.46
May-97	45.47	49.49	43.86	47.77	61.97	39.15
Jun-97	44.97	49.85	43.62	47.44	62.09	38.86
Jul-97	45.41	50.13	44.17	47.80	62.15	39.49
Aug-97	45.88	50.42	45.24	48.04	62.34	40.67
Sep-97	45.86	50.73	45.31	48.09	62.30	40.76
Oct-97	46.09	51.03	44.96	48.52	62.22	40.37
Nov-97	46.29	51.33	45.69	48.59	62.24	41.22
Dec-97	46.16	51.61	46.28	48.37	62.29	41.88
Jan-98	45.79	51.87	45.62	48.28	62.36	41.11
Feb-98	46.00	52.07	46.09	48.40	62.30	41.66
Mar-98	46.92	52.39	46.26	49.38	62.34	41.84
Apr-98	47.68	52.82	47.32	49.96	62.50	43.04
May-98	48.37	53.17	47.52	50.70	62.60	43.24
Jun-98	48.47	53.50	49.13	50.41	62.65	45.08
Jul-98	48.65	53.78	48.46	50.88	62.84	44.29
Aug-98	48.98	54.11	48.75	51.23	62.99	44.59
Sep-98	48.90	54.26	49.69	50.91	65.17	45.28
Oct-98	49.42	54.46	50.34	51.29	66.37	45.81
Nov-98	49.77	54.70	53.05	50.88	66.48	48.90
Dec-98	49.58	55.01	53.17	50.74	67.17	48.91
Dec-98	48.87	55.23	50.39	50.96	67.42	45.67
Feb-99	49.22	55.41	51.30	51.09	68.35	46.55
Mar-99	49.21	55.61	50.80	51.29	68.81	45.89
Apr-99	49.52	56.03	49.82	52.04	68.89	44.75
May-99	49.95	56.19	49.50	52.62	68.97	44.37
Jun-99	49.40	56.40	49.60	52.11	68.99	44.48
Jul-99	49.84	56.61	49.70	52.57	69.03	44.59
Aug-99	50.10	56.77	49.94	52.82	69.03	44.86
Sep-99	49.94	56.96	49.88	52.74	69.06	44.79
Oct-99	50.73	57.14	50.21	53.48	70.51	44.91
Nov-99	50.63	57.33	50.60	53.33	70.57	45.34
Dec-99	50.23	57.48	50.46	53.01	70.66	45.16
Jan-00	50.17	57.70	50.15	53.11	70.73	44.79
Feb-00	50.01	57.89	50.02	53.05	70.82	44.63
Mar-00	50.45	58.12	50.37	53.46	70.92	45.02
Apr-00	51.16	58.71	50.83	54.21	71.00	45.53
May-00	51.30	59.11	49.29	54.96	71.07	43.75
Jun-00	50.42	59.22	51.43	53.44	71.10	46.20
Jul-00	52.33	59.32	52.42	55.09	68.68	47.78
Aug-00	51.98	59.47	51.84	54.97	69.78	46.91
Sep-00	52.38	59.71	52.93	55.11	69.43	48.22
Oct-00	52.35	59.91	52.26	55.35	69.50	47.44
Nov-00	52.79	60.12	54.05	55.31	69.59	49.49
Dec-00	51.98	60.18	51.79	55.20	69.64	46.88
Jan-01	52.18	60.37	52.73	55.18	69.78	47.93
Feb-01	52.04	60.49	52.02	55.29	69.76	47.12
Mar-01	52.35	60.65	52.63	55.47	69.77	47.82

Table 6, continued
Time eriod	GCPI	SCPI	TCPI	NTCPI	TCPI_NFNE	TCPI_FE
Apr-01	52.90	61.00	52.37	56.20	69.79	47.52
May-01	52.92	61.28	50.82	56.79	69.82	45.73
Jun-01	52.63	61.46	51.87	56.23	69.83	46.94
Jul-01	53.44	62.40	52.67	57.09	70.90	47.66
Aug-01	53.83	62.60	53.07	57.41	71.22	48.09
Sep-01	53.91	62.67	53.40	57.42	71.34	48.47
Oct-01	54.17	62.75	53.65	57.61	71.52	48.74
Nov-01	54.39	62.78	53.83	57.77	71.46	48.99
Dec-01	53.97	62.88	53.62	57.50	71.52	48.70
Jan-02	54.10	62.96	53.21	57.75	71.57	48.16
Feb-02	54.16	63.01	53.80	57.66	71.82	48.85
Mar-02	55.39	63.05	55.39	58.34	71.90	50.84
Apr-02	55.77	63.12	55.78	58.59	72.03	51.30
May-02	54.55	63.40	54.15	58.06	72.57	49.09
Jun-02	55.25	63.45	54.69	58.56	73.12	49.62
Jul-02	56.31	63.59	56.35	59.10	73.18	51.72
Aug-02	56.55	63.63	56.95	59.18	73.21	52.48
Sep-02	56.71	63.66	57.39	59.21	72.95	53.10
Oct-02	56.05	63.85	56.18	59.02	72.87	51.59
Nov-02	56.67	63.97	56.07	59.64	74.03	51.13
Dec-02	56.06	64.13	55.50	59.32	75.23	50.08
Jan-03	56.45	64.02	56.07	59.47	75.98	50.60
Feb-03	56.96	64.10	57.22	59.64	76.11	52.03
Mar-03	56.87	64.39	56.64	59.82	76.05	51.30
Apr-03	57.03	64.75	56.08	60.25	76.23	50.55
May-03	56.76	64.77	55.64	60.14	76.65	49.87
Jun-03	56.52	64.99	56.12	59.89	76.65	50.47
Jul-03	57.28	65.14	57.51	60.25	76.57	52.28
Aug-03	57.73	65.36	58.07	60.58	76.66	52.95
Sep-03	58.14	65.58	57.57	61.16	76.80	52.28
Oct-03	59.29	65.80	59.43	61.76	76.80	54.66
Nov-03	59.67	66.15	60.63	61.91	76.83	56.17
Dec-03	60.29	66.46	61.17	62.43	77.15	56.76
Jan-04	60.07	66.78	61.01	62.40	77.35	56.49
Feb-04	59.57	67.22	60.00	62.40	77.25	55.26
Mar-04	60.21	67.82	59.29	63.38	76.03	54.68
Apr-04	60.70	68.37	60.15	63.80	74.84	56.10
May-04	61.11	68.96	59.34	64.60	74.91	55.07
Jun-04	61.83	69.46	60.11	65.22	74.97	56.04
Jul-04	62.88	69.98	61.12	66.08	74.94	57.33
Aug-04	63.19	70.51	61.59	66.43	75.09	57.89
Sep-04	63.39	70.86	62.02	66.63	75.88	58.19
Oct-04	64.26	71.38	62.82	67.38	76.17	59.15
Nov-04	65.09	71.93	64.15	67.97	76.52	60.73
Dec-04	64.05	72.51	62.10	67.82	76.85	58.03
Jan-05	64.71	73.09	63.41	68.29	76.69	59.72
Feb-05	65.33	73.74	63.83	68.97	76.77	60.25
Mar-05	66.30	74.32	65.16	69.69	76.98	61.87
Apr-05	67.56	75.16	66.09	70.88	76.97	63.08
May-05	66.81	75.82	64.50	70.89	77.27	61.00
Jun-05	66.73	76.34	65.22	70.83	77.34	61.92
Jul-05	68.07	76.82	65.75	72.05	77.43	62.54
Aug-05	67.87	77.33	65.99	72.01	77.50	62.82
Sep-05	68.16	77.85	66.49	72.34	77.54	63.43
Oct-05	68.54	78.69	66.57	72.97	77.76	63.50
Nov-05	69.31	79.11	67.70	73.51	77.79	64.90
Dec-05	68.83	79.55	67.04	73.43	78.03	63.98
Jan-06	69.81	79.96	68.68	74.02	78.36	65.96
Feb-06	70.00	80.30	68.05	74.49	78.60	65.12

Table 6, continued
Time eriod	GCPI	SCPI	TCPI	NTCPI	TCPI_NFNE	TCPI_FE
Mar-06	70.12	80.59	67.66	74.81	78.90	64.53
Apr-06	70.86	81.19	68.90	75.36	79.30	66.01
May-06	71.03	81.76	68.47	75.83	80.50	65.19
Jun-06	71.52	82.15	68.98	76.28	79.67	66.08
Jul-06	72.97	82.60	70.65	77.29	80.06	68.07
Aug-06	74.06	83.16	71.66	78.20	80.13	69.33
Sep-06	74.18	83.73	72.04	78.42	79.99	69.82
Oct-06	74.38	84.13	72.08	78.74	79.81	69.97
Nov-06	75.05	84.43	74.26	78.87	80.33	72.55
Dec-06	75.44	84.72	77.19	78.55	80.56	76.19
Jan-07	74.32	85.08	74.01	78.55	80.90	72.04
Feb-07	75.25	85.56	74.42	79.45	82.85	72.06
Mar-07	75.81	85.54	73.73	80.11	84.40	70.74
Apr-07	75.99	85.89	73.31	80.52	84.66	70.16
May-07	76.82	86.26	75.71	80.75	84.86	73.22
Jun-07	76.82	86.99	76.38	80.86	85.15	74.02
Jul-07	77.65	87.74	76.00	81.97	85.49	73.39
Aug-07	78.89	88.31	77.13	82.98	85.70	74.77
Sep-07	81.11	88.89	79.42	84.55	86.67	77.38
Oct-07	82.33	89.43	81.21	85.36	87.21	79.57
Nov-07	82.28	90.01	81.20	85.55	88.55	79.13
Dec-07	82.72	90.68	80.36	86.41	88.69	78.01
Jan-08	84.66	91.46	80.70	88.33	89.71	78.14
Feb-08	84.92	92.25	84.44	87.87	91.39	82.47
Mar-08	88.35	93.15	88.77	90.08	93.75	87.30
Apr-08	91.31	95.06	89.54	93.22	94.27	88.17
May-08	94.08	96.64	91.54	95.73	94.93	90.65
Jun-08	96.46	97.84	95.02	97.38	95.77	94.91
Jul-08	100.00	100.00	100.00	100.00	100.00	100.00
Aug-08	102.24	101.03	102.23	101.67	101.47	102.94
Sep-08	103.29	101.69	102.10	102.96	101.40	102.72
Oct-08	105.57	102.30	102.00	105.31	101.18	102.67
Nov-08	105.22	102.44	98.72	106.25	96.88	100.23
Dec-08	104.69	102.87	98.86	105.92	97.22	100.19
Jan-09	103.88	104.99	97.11	106.82	97.65	96.48
Feb-09	105.62	105.50	98.99	107.93	98.51	99.27
Mar-09	106.10	105.08	99.68	107.92	99.28	99.88
Apr-09	106.90	105.93	99.73	109.01	99.67	99.59
May-09	106.13	106.45	97.02	109.54	98.60	95.42
Jun-09	107.19	107.02	98.85	110.10	100.59	97.12
Jul-09	108.98	108.08	101.51	111.24	99.90	102.77
Aug-09	111.96	108.73	103.37	113.52	102.23	104.24
Sep-09	112.43	109.06	103.78	113.94	101.81	105.39
Oct-09	113.64	109.51	104.28	115.05	101.98	106.16
Nov-09	115.94	110.10	107.49	116.23	103.99	110.45
Dec-09	114.65	110.42	106.95	115.42	104.76	108.75
Jan-10	118.59	111.44	108.58	118.82	107.21	109.68
Feb-10	118.60	112.31	107.90	119.48	106.67	108.83
Mar-10	119.97	113.47	109.41	120.70	107.14	111.29
Apr-10	122.02	115.05	114.17	121.56	110.23	117.57
May-10	121.50	115.44	112.51	121.88	108.90	115.61
Jun-10	121.74	116.01	111.97	122.55	108.29	115.11
Jul-10	124.66	117.82	114.07	125.24	109.83	117.72
Aug-10	128.81	118.11	119.55	127.11	110.30	127.70
Sep-10	132.69	118.26	123.08	129.37	110.51	134.21
Oct-10	134.33	120.24	124.11	131.38	113.58	133.42
Nov-10	136.53	120.72	129.72	131.55	114.80	142.98
Dec-10	135.58	121.51	128.96	131.36	115.80	140.64
Jan-11	137.01	122.85	130.84	132.59	116.72	143.39
Feb-11	135.56	123.23	127.33	132.73	117.33	136.19

Table 6, continued
Time eriod	GCPI	SCPI	TCPI	NTCPI	TCPI_NFNE	TCPI_FE
Mar-11	137.70	123.73	129.00	134.27	119.53	137.41
Apr-11	139.27	125.26	132.01	135.30	123.70	139.42
May-11	138.76	125.62	131.07	135.36	125.44	136.10
Jun-11	140.76	125.97	131.37	137.19	126.13	136.06
Jul-11	142.78	127.71	133.59	139.01	126.08	140.28
Aug-11	145.77	128.25	135.40	141.28	128.54	141.52
Sep-11	147.65	128.56	136.69	142.63	130.70	142.01
Oct-11	150.42	130.73	141.11	144.53	131.83	149.38
Nov-11	151.21	131.25	142.99	144.80	132.44	152.40
Dec-11	149.56	131.64	141.17	144.17	133.45	148.06
Jan-12	152.53	133.13	143.80	146.57	135.33	151.36
Feb-12	152.60	133.94	142.92	147.33	137.60	147.69
Mar-12	154.43	134.62	145.89	148.20	141.08	150.26
Apr-12	156.96	136.78	148.83	150.42	143.52	153.66
May-12	158.82	137.18	147.54	152.72	143.96	150.85
Jun-12	158.70	138.01	146.44	153.39	142.41	150.11
Jul-12	156.28	139.76	147.78	151.58	140.81	154.06
Aug-12	159.04	140.03	151.77	152.74	144.31	158.54
Sep-12	160.48	140.24	152.31	153.92	147.97	156.33
Oct-12	161.41	142.03	153.61	155.12	149.10	157.79
Nov-12	160.81	141.91	154.04	154.38	149.91	157.86
Dec-12	161.42	142.19	154.53	154.88	151.67	157.25
Jan-13	163.95	144.50	156.02	157.68	152.93	158.96
Feb-13	162.71	145.01	154.71	157.28	153.56	155.91
Mar-13	163.07	145.59	156.20	157.34	153.57	158.71
Apr-13	164.06	147.18	158.03	158.31	154.47	161.37
May-13	163.87	147.43	156.54	158.79	153.62	159.27
Jun-13	165.95	147.71	156.94	160.63	155.66	157.95
Jul-13	169.59	150.34	163.52	162.74	157.53	168.75
Aug-13	172.85	150.90	165.69	165.12	159.49	171.12
Sep-13	172.13	151.50	163.78	165.45	161.77	165.45
Oct-13	177.06	153.20	167.83	169.19	163.45	171.54
Nov-13	180.58	153.43	175.29	169.75	164.40	184.87
Dec-13	177.27	153.69	169.01	169.18	164.57	172.76
Jan-14	177.12	155.72	167.04	170.71	165.43	168.27
Feb-14	175.92	155.92	164.84	170.52	165.85	163.70
Mar-14	177.89	156.10	167.76	171.31	166.00	169.07
Apr-14	179.35	159.92	171.31	173.12	166.39	175.46
May-14	177.76	160.54	168.96	172.84	167.36	170.14
Jun-14	179.37	161.05	171.37	173.65	168.16	173.99
Jul-14	182.51	163.18	176.13	175.74	169.37	181.93
Aug-14	183.91	163.40	176.22	177.06	169.84	181.70
Sep-14	184.53	163.81	176.56	177.68	169.99	182.20
Oct-14	185.24	165.00	177.42	178.56	169.59	184.18
Nov-14	184.83	164.82	175.27	178.88	167.83	181.60
Dec-14	182.41	164.81	171.17	178.19	165.81	175.59
Jan-15	181.44	166.54	169.73	178.65	164.48	173.98
Feb-15	178.87	166.34	165.93	177.63	162.35	168.64
Mar-15	179.05	166.37	166.78	177.50	162.38	170.22
Apr-15	180.33	169.20	168.51	179.34	164.30	171.82
May-15	180.73	169.32	168.59	179.73	164.40	171.87
Jun-15	183.09	169.49	171.74	180.78	165.57	176.80
Jul-15	183.47	170.87	171.90	181.71	165.82	176.88
Aug-15	184.65	170.95	172.70	182.50	165.83	178.37
Sep-15	185.23	171.10	173.44	182.82	165.24	180.28
Oct-15	186.65	172.42	175.92	183.82	165.76	184.52
Nov-15	188.45	172.56	179.53	184.20	166.60	190.63
Dec-15	187.22	172.76	178.52	183.56	167.20	188.16
Jan-16	186.48	174.91	176.72	184.55	167.69	184.32
Feb-16	185.87	175.23	176.06	184.40	166.71	183.91
Mar-16	186.27	175.23	175.66	184.89	165.41	184.31

Table 6, continued
Time eriod	GCPI	SCPI	TCPI	NTCPI	TCPI_NFNE	TCPI_FE
Apr-16	187.73	177.72	176.87	186.93	166.61	185.52
May-16	185.72	178.04	173.86	186.36	166.83	179.63
Jun-16	187.83	178.27	177.12	187.18	167.13	185.53
Jul-16	189.88	179.59	179.72	188.70	167.83	189.84
Aug-16	190.09	179.78	179.67	188.99	168.37	189.27
Sep-16	190.96	180.53	179.89	190.03	168.23	189.83
Oct-16	192.71	182.85	180.30	192.53	168.65	190.23
Nov-16	194.04	183.10	183.41	192.72	169.45	195.40
Dec-16	191.83	183.24	183.11	190.93	170.50	193.89
Jan-17	191.42	185.39	181.24	192.24	170.97	189.93
Feb-17	192.31	185.60	182.61	192.64	172.05	191.56
Mar-17	193.96	185.77	184.32	193.58	172.51	194.38
Apr-17	194.61	189.16	185.69	195.25	173.34	196.25
May-17	193.19	189.46	183.64	194.86	173.67	192.06
Jun-17	192.05	189.61	179.78	195.30	173.78	184.65
Jul-17	192.11	191.33	179.13	196.39	173.84	183.35
Aug-17	193.38	191.41	180.81	196.96	173.84	186.53
Sep-17	196.06	191.53	186.07	197.52	174.88	195.58
Oct-17	197.75	193.99	188.32	199.36	175.50	199.30
Nov-17	199.43	194.55	190.95	200.18	177.03	202.93
Dec-17	198.39	194.76	188.08	200.38	178.21	196.44
Jan-18	198.25	196.33	186.23	201.65	180.69	190.78
Feb-18	197.38	196.63	184.81	201.53	181.74	187.13
Mar-18	196.74	198.18	183.47	202.17	183.75	183.05
Apr-18	198.63	203.12	186.20	205.18	185.51	185.00
May-18	198.37	203.50	185.02	205.55	187.28	182.12
Jun-18	199.45	204.27	186.09	206.49	189.05	182.85

Higher inflation rate in non-tradable products prices is usually called a signal for increased profitability in the non-traded sector and thus encouraging recourses inflows to domestic sector. We would like to interpret this as an ‘internal competitiveness.’ This ‘internal competitiveness’ could be one of the reasons for low trade to GDP ratio of Pakistan economy. It needs further and deeper analysis.

8. Limitations of the study and future research direction

Due to limitations of data, results cannot be as accurate as could have been had we production data for those few goods for which we have used judgement. Moreover, results could have been improved if we could have estimated trade shares for each of the year rather than for few selected years. We expect these indices will be free from such shortcomings in case PBS starts compiling and disseminating such indices on regular basis.

We think there is also need to classify output for tradable and non-tradable sectors for the case of Pakistan as has already been done for some countries like Australia by Knight et al. [7]. Now after having the dataset for tradable and non-tradable products’ prices in Pakistan, there should be some work on exploring the source of differentials in tradable and non-tradables’ inflation in the country beyond the simple correlation coefficient of tradables and exchange rate of PKR we analysed to surmise that tradables face more competition. There could be differences in productivity in tradable and non-tradable sectors (the supply side factor) or there could be some shifts in demand responsible for the observed dynamics of tradable to non-tradable prices ratio. Moreover, in view of rising total imports of Pakistan (as of fiscal year 2018), it would be interesting if we estimate an import demand function for the country by considering tradable prices separately.

9. Summary and conclusion

Pakistan Bureau of Statistics collects, compiles, and disseminates monthly data on consumer price index (CPI) as well as indices for selected few subgroups within the CPI basket. Economic analysts, researchers and policy makers also need such price indices for tradable and nontradable sectors (of the economy) to analyze inflation rate dynamics in depth for the case of Pakistan. These type of indices are available for developed countries, like OECD countries, to investigate various international macroeconomic dynamics pertaining to these countries. There is no such dataset for the case of Pakistan. This study is an attempt to fill this gap.

This study, at first stage, divided CPI basket into goods and services and considered most of the services to be non-tradables following Johnsons (2017) suggestion that services should be treated as non-tradable (due to their immobility). After this classification, items in the CPI basket were bifurcated as tradable/non-tradables. This study selected mostly used bifurcation criterion i.e. trade to output ratio (TOR) with a threshold of 10 percent for goods to be classified as tradables. Any item in the CPI basket for which TOR is greater than or equal to 10 percent in Pakistan is treated as a tradable one. Based on this bifurcation we estimated the tradable and nontradable price indices for Pakistan for the period of July 1991 to June 2018. Estimated ‘tradable prices sub-index’ is also crosschecked with weighted average of unit value indices of exports and imports of the country. Tradables’ price indices for food and energy subgroup and for nonfood-non-energy subgroup are also estimated in this study.

Our findings suggest that share of tradable items in CPI basket has increased from one-fifth in early 1990s to about one-fourth recently. Thus, three-fourths of the overall CPI basket are non-tradables. Out of the one fourth tradable items in CPI basket, only 6.6 percent are non food non-energy goods.

Based on some basic statistics, we can infer that YoY inflation rate in tradable sector is lower but volatile compared to YoY inflation rate in nontradable sector. Since higher YoY inflation in nontradables may guide resource allocation towards nontradable sector, Pakistan can be interpreted as an ‘internally competitive’ economy. This could be one of the major reasons of chronic trade deficits of Pakistan.

Higher volatility in tradable sector inflation rate compared to nontradable sector inflation rate is also vindicated by the behaviour of inflation diffusion indices of these subgroups. Negative values of inflation diffusion index for tradable goods suggest that at times number of goods for which prices fall are more than the number of goods for which prices increase during a given month. This characteristic is hardly observed in case of nontradable goods’ prices, notwithstanding the fact that prices do fall for nontradables as well. Moreover, tradable sector inflation causes inflation in non-tradeable sector of the country. Movement in tradable goods inflation rate in Pakistan is observed to be associated with foreign exchange rate dynamics [of national currency – Pak Rupee – vis-a-vis US dollar]; wherein we observe what we call ‘Human Resources induced Dutch Disease’. We could not find significant result for persistence in tradable goods inflation in Pakistan. Lack of persistence observed in tradable prices’ month on month inflation suggests that there is a tendency in tradables inflation to converge ‘quickly’ to its long run level following any shock.

The tradable and nontradable price indices estimated in this study are going to be useful for a) answering the questions like relative performance of tradable and nontradable sectors in the economy, b) exploring the determinants of tradable and nontradable inflation in Pakistan, c) estimating difference in exchange rate pass through to tradable and nontradable inflation and d) investigating the international trade competitiveness of Pakistan.

Footnotes

See Pakistan Bureau of Statistics [].

For 2007–08 basket these are: “Food and Non-Alcoholic Beverages”, “Alcoholic Beverage and Tobacco”, “Clothing and Footwear”, “Housing, Water, Electricity, Gas and Other Fuels”, “Furnishing and Household Equipment Maintenance”, “Health and Transport”, “Communication”, “Recreation and Culture”, “Education”, and “Restaurants and Hotels”. Other goods and services which do not come under any of these groups are classified as “Miscellaneous Goods and Services”.

We propose PBS to prepare and disseminate these new sub-indices within CPI basket for general consumption of stakeholders including economic analysts and researchers.

Since 1947, PBS has prepared 7 baskets of CPI. We could not find commodity wise prices of CPI basket prior to 1991 readily available.

For example, ‘butter locally packed’.

For example, ‘Ponstan tablet’.

For example, ‘mobile phones’.

95 percent of this range is just $-$ 0.36 to 0.59.

95 percent of this range is just $-$ 0.26 to 0.19. We expect there will be no difference when PBS would start releasing such sub-indices in future, as we propose in this study, with the introduction of new base year for price data.

RHS in the parenthesis of this variable’s name AUXM in the graph means (y-axis) scale for this variable is right hand side (RHS).

Before Granger non-causality analysis we tested both the inflation rate series for unit root and found them to be stationary.

Acknowledgments

Authors would like to thank Leena Hietaniemi, Rafael Posse-Fregoso, Ali Choudhary, Sajawal Khan, Javed Iqbal and Adila Ayaz for their valuable comments that helped improve this study.

References

Hanif

Iqbal

Khan

. Global commodity prices and domestic inflation: A case study of Pakistan. SBP Research Bulletin. 2017; 13(1): 21-51.

Hanif

Batool

. Openness and inflation: A case study of Pakistan. MPRA Working Paper No. 10214. Available: https//mpra.ub.uni-muenchen.de/id/eprint/10214, 2006.

Hanif

. A note on food inflation in Pakistan. Pakistan Economic and Social Review. 2012; 50(2): 183-206.

Hanif

Iqbal

. Assessing monetary policy effectiveness in rich data environment. SBP Working Paper No. 80, State Bank of Pakistan, 2016.

Pakistan Bureau of Statistics. Methodology of price collection and computing price indices, 2012.

Dwyer

. The tradeable non-tradeable dichotomy: a practical approach. Australian Economic Papers. 1992; 31(59): 443-459.

Knight

Johnson

. Tradables: developing output and price measures for Australia’s tradable and non-tradable sectors. Australian Bureau of Statistics: Australia, 1997.

Salter

. Internal and external balance: the role of price and expenditure effects. Economic Record. 1959; 35(71): 226-238.

Goldstein

Officer

. New measures of prices and productivity for tradable and nontradable goods. Review of Income and Wealth. 1979; 25(4): 413-427.

10.

De Gregorio

Giovannini

Wolf

. International evidence on tradables and nontradables inflation. European Economic Review. 1994; 38(6): 1225-1244.

11.

Dixon

Griffiths

Lawson

. Exploring tradable and non-tradable inflation in consumer prices. Paper presented at the New Zealand Association of Economists Conference. Wellington: New Zealand, 2004.

12.

Amador

Soares

. Competition in the Portuguese economy: a view on tradables and non-tradables. Economic Bulletin and Financial Stability Report Articles, 2012, pp. 41-57.

13.

Attewell

Crossan

. The tradable sector and its relevance to New Zealand’s GDP. Paper presented at the New Zealand Association of Economists Conference. Wellington: New Zealand, 2013.

14.

Canas

Gouveia

. Allocation of resources between the tradable and non-tradable sectors: developing a new identification strategy for the tradable sector. GEE Papers. 2016; 60: 1-15.

15.

Johnson

. Tradable and nontradable inflation indexes: replicating New Zealand’s tradable indexes with BLS CPI data. Monthly Labor Review. 2017; 140(1): 1-25.

16.

Hanif

Malik

Iqbal

. Intrinsic inflation persistence in a developing country. SBP Research Bulletin. 2016; 12(1): 19-42.

17.

Hanif

Khan

Ghouri

. Inflation Monitor. Research Department, State Bank of Pakistan, 2011, p. 42.

18.

Engel

. Real exchange rates and relative prices: An empirical investigation Journal. of monetary economics. 1993; 32(1): 35-50.

19.

Choudhary

Naeem

Faheem

Hanif

Pasha

. Formal sector price discoveries: preliminary results from a developing country. MPRA Working Paper No. 32368, Available: https//mpra.ub.uni-muenchen.de/id/eprint/32368, 2011.

Tradable and non-tradable inflation in consumer prices in Pakistan 1

Abstract

Keywords

1. Introduction

2. Literature review

3. Motivation

4. Conceptual framework

Table 1 List of new sub-indices estimated for Pakistan in this study

7. Results and discussion

Table 2 Number (percent share) of items in new sub-groups

Table 4 Some basic and advanced statistics on MoM change in sub-indices within CPI

9. Summary and conclusion

Footnotes

Acknowledgments

References

Table 1
List of new sub-indices estimated for Pakistan in this study

Table 2
Number (percent share) of items in new sub-groups

Table 4
Some basic and advanced statistics on MoM change in sub-indices within CPI