Retail Food Price Inflation Modelling Project

Retail Food Price Inflation Modelling Project

Final Report

T.A. Lloyd, C.W. Morgan

(University of Nottingham)

J. Davidson, A. Halunga,S. McCorriston

(University of Exeter)

28th April 2011

Contents

Executive Summary1

Aims of the Study2

Section 1: UK Consumer Food Price Inflation4

Section 2: The Drivers of UK Food Price Inflation12

Section 3: Modelling UK Consumer Food Price Inflation27

Section 4: Forecasting and Predictions46

Section 5: Conclusions51

Appendices

References
Listing of the Fitted Model
Within Sample Forecasts of Food Sub-group Models
Data definitions and sources

Executive Summary

Food price inflation peaked in the summer of 2008 at nearly 15%, a level not seen for several decades. While the causes of this spike in food prices were of great interest, a more pressing concern was to focus on the longer-term drivers of retail food prices, particularly with regard to being able more readily to predict how they might develop in the future. The focus on prediction lies at the heart of the project commissioned by DEFRA in the winter of 2009/10 and for which this report forms the outcome of the research undertaken.

Food price inflation has been rising at a time when world food and other commodity prices - for example, wheat, oil and copper - have reached relatively high levels and it is possible to draw a simple conclusion that domestic food price inflation simply reflects changing world commodity market conditions. The report presented here demonstrates that while there is some merit in this argument, it is in fact a much more complex process than this line of argument suggests.

Empirical analysis based on modern econometric techniques shows that the major drivers of the UK food price inflation, as measured by the CPI for food, are world commodity prices, the Dollar-Sterling exchange rate, unemployment, labour costs and the price of oil. We find that food price inflation is relatively unresponsive to changes in these drivers. The most important determinant of food price inflation is world food prices with exchanges rates also exerting a significant effect. We also find an important indirect role for oil through world food commodity prices. These results imply that large changes in the values of these drivers are required to affect domestic food price inflation. Results also suggest that shocks that persist will tend to have a larger impact on food price inflation than one-off shocks since the effect accumulates over time.

The estimated model forms the basis for a forecasting tool which not only provides an opportunity to test different scenarios in a “What if...?” manner but also delivers monthly forecasts of the level of food price inflation with appropriate bounds of confidence applied in each case.

Aims of the Study

The boom in the prices of commodities in 2007 and 2008 led to a spike in inflation in many economies across the globe that was quite sharp and somewhat unexpected. While the response of the price of manufactured goods to rises in industrial commodity prices (for example, oil) was significant, perhaps of more concern for all economies was the way consumer food prices responded to sharp increases in soft commodity prices in a relatively short period of time.

Although world commodity prices have subsequently fallen back, the emerging consensus is that commodity prices will remain at higher levels relative to the past two decades and be more volatile. The obvious concern that arises is the impact on households, particularly in countries where expenditure on food accounts for a relatively high share of total consumer expenditure. Even in countries where food expenditure accounts for a relatively small share of the household expenditure, such as the UK, there are still important distributional issues as the impact on the poorer sections of society can be considerable even though the aggregate effect is small.

The recent developments in world commodity markets and the impact on consumer prices have also raised concerns about future food security in many countries, including the UK. Given the weight food products have in the calculation of consumer price inflation in many countries, understanding the factors that caused this inflation is important but perhaps more informative for policy makers is the ability to forecast any future rises with greater accuracy and confidence.

The research project therefore has three main aims which are:

To review and evaluate existing understanding of UK consumer food price inflation
To model the drivers of UK consumer food price inflation
To create a tool that allow for forecasting and scenario setting to help policy makers understand with greater certainty how UK consumer food price inflation might develop in the future.

The research presented in this report outlines how these three aims were met. The work was undertaken with a view to developing a forecasting tool that can allow policy makers to undertake scenario setting exercises so that future patterns of food price inflation could be understood with greater certainty than previously. As such, the report is structured as follows. In Section 1, we review the history of consumer food price inflation in the UK and make benchmark comparisons with other OECD countries. In Section 2, we provide the background literature to the issue of food price inflation, covering the main causes of the 2007-2008 commodity price spike, the literature on price transmission and, more generally, the links between commodity markets and inflation. This background serves as the basis for selecting the appropriate econometric methodology which is outlined in Section 3; this provides details on the rationale for the econometric methodology chosen, the basis of the specific econometric model and results relating to the long-run relationship between the main drivers of domestic retail food prices and, in particular, the links between domestic food price inflation and world commodity markets. This econometric model also serves as the basis for the forecasting tool, the interpretation of which is provided in Section 4. Section 5 concludes,

In achieving these aims, the report begins by reviewing the evidence contained in food price inflation data in the UK and then subsequently the views of the academic literature that identify key drivers in food price inflation. Then drawing on this, data are collected from publically available sources on a monthly basis to which modern time-series econometric techniques are applied. The econometric analysis will focus on a number of issues pertinent to food price inflation namely: magnitude and dynamic response to shocks in the drivers of inflation and the extent of potential interaction between drivers. The primary focus will be on aggregate food price inflation, as recorded in the Consumer Price Index (CPI), with additional special analyses of specific food products. The outcomes from the research will be the development of a clearer understanding of the factors that drive food price inflation, quantification of the effects of the drivers and the dynamic response of food price inflation to these drivers, understanding of price transmission mechanism in specific food products and the creation of a tool to forecast food price inflation for subsequent use by DEFRA.

Section 1: UK Consumer Food Price Inflation

The main measure used to calculate inflation in the UK is the Consumer Price Index (CPI), which became the preferred measure in December 2003 replacing the Retail Price Index (RPI). Both are still calculated although differences in the recorded values reflect the items included in each index[1]; the major difference between the two is that the RPI includes the costs of the housing market (mortgage interest for example) whereas the CPI does not. For the purposes of this study we will be focussing mainly on the CPI and in particular the component relating to Food and Non-Alcoholic Beverages. Divisions within the CPI are given a weighting to reflect their importance within the consumer basket as shown in Table 1. The weights reflect spending in each specific year while the relative importance of food in the overall CPI measure have increased in recent years as food prices have risen.

Table 1: Divisions and Weights within the CPI (Source: ONS, 2007)

Divisions / Weight
01 Food and Non-Alcoholic Beverages / 103
02 Alcoholic Beverages and Tobacco / 43
03 Clothing and Footwear / 62
04 Housing, Water, Electricity, Gas and other Fuels / 115
05 Household Furnishings, Equipment and Maintenance / 68
06 Health / 24
07 Transport / 152
08 Communications / 24
09 Recreation and Culture / 153
10 Education / 18
11 Restaurants and Hotels / 138
12 Miscellaneous Goods and Services / 100

All the weights add to 1000 and the largest two categories are Recreation and Culture (153) and Transport (152). The weight given to Food and Non-Alcoholic Beverages is just over 10% of the total index (103) and is thus a key component of the measure of inflation each month. All indices are calculated from samples of prices of goods and services and in effect both indices (the RPI and the CPI) are “an average measure of change in the prices of goods and services bought for the purpose of consumption by the vast majority of households in the UK” (ONS, 2007).

Figure 1 plots the index for all goods (UKCPI), the index for food (UKFCPI) and the index for non-food items (UKNONFCPI), over a 22 year period. It is noticeable that while the trends are similar there are divergences between the two, including most notably from 2007 when the food index rises at a much faster rate than the measures for all goods and non-food items.

Figure 1: All Goods, Food and Non-Food CPI 1988(1) – 2010(1) (2005 = 100)

Source: ONS

The Food and Non-Alcoholic Beverages category itself is sub-divided into a number of headings too. Table 2 shows that the category is split such that the weight for Food is 90 and for Non-Alcoholic Beverages it is 13. Most importance is attached to Meat (21), Bread and Cereals (15) and Vegetables including Potatoes and Tubers (14).

Table 2: Groups, Classes and Weights within Food Non-alcoholic Beverages

Groups and Classes / Weight
01.1 Food / 90
01.1.1 Bread and Cereals / 15
01.1.2 Meat / 21
01.1.3 Fish / 4
01.1.4 Milk. Cheese and Eggs / 12
01.1.5 Oils and Fats / 2
01.1.6 Fruit / 9
01.1.7 Vegetables including potatoes and tubers / 14
01.1.8 Sugar, jam, syrups, chocolate and confectionery / 11
01.1.9 Food products n.e.c. / 2
01.2 Non alcoholic beverages / 13
01.2.1 Coffee, tea and cocoa / 3
01.2.2 Mineral waters, soft drinks and juices / 10

Source: ONS (2007)

The CPI is calculated and reported each month based on the sample prices collected. However, this in itself is not a measure of inflation. Inflation is defined as the annual change in the CPI and so to provide a numeric value for inflation each month, the index must be compared to its value 12 months previously. Thus, for example, if the index was calculated to be 110 in March 2010 and 100 in March 2009 then the annual rate of inflation in March 2010 is deemed to be 10% as the index has risen by 10% when comparing those two months. The important point is the comparison to a year ago as prices for specific products and categories might have fallen from one month to the next – suggesting deflation in the mind of the consumer - but in relation to the same time one year ago, they could well have risen and thus inflation is deemed to be rising.

Figure 2 plots inflation as measured by the annual change in the UKCPI, the CPI for Food and Non-alcoholic Beverages (UKCPI) and for non-food items (UKNONFCPI).

Figure 2: Annual Inflation for All Goods, Food and Non-food CPI 1989(1)-2010(1) (%)

Source: (ONS)

Again, it is apparent that while the all items and non-food indices move relatively closely together apart from some disparity in the early 1990s, the major difference lies when we consider the food index against the other two. Here there is much greater volatility including periods when the index goes negative – i.e. the price of food items was actually falling in nominal terms in 1997, 2000, 2002, 2005 and 2006. Equally, there are significant peaks in food price inflation in July 1995 and mid-2001 as well as the outlier spike in prices already mentioned in 2008. The evidence would suggest that food price inflation does indeed behave differently to non-food price inflation and thus understanding the drivers of food price inflation becomes a highly specific activity.

The inflation rates of the various sub-components of the Food and Non-Alcoholic Beverages index behave in different ways over time as shown in Figures 4-6. In these figures, we utilise RPI data as they have a longer time series and thus help provide a longer-term picture of what has happened in these sub-groups. Some of the highest levels of inflation can be found in Vegetables and Potatoes (Figure 3) where a peak of over 60% can be observed in 1975, with Milk & Eggs and Butter also reached over 40% at the same time (Figure 4), although the outstanding level is that of sugar which hit an inflation level of over 80% (Figure 5). 1975 was a drought year in the UK and could possibly explain some of the very high prices experienced in that year along with world sugar markets being heavily distorted too.

Figure 3: Annual Inflation for Bread, Meat and Vegetables 1957(1) – 2010(1)

Source: ONS

Figure 4: Annual Inflation for Butter and Milk & Eggs 1957(1) – 2010(1)

Source: ONS

Figure 5:Annual Inflation for Sugar and Fruit 1957(1) – 2010(1)

SSource: ONS

It is evident that patterns of food price inflation are very different across the sub-categories of the food index. There are often periods of negative inflation – something rarely seen in the headline all items inflation index although the RPI in 2009 did turn negative for several months – and there is a degree of volatility that perhaps could reflect a range of supply and demand factors that influence the prices consumers pay for their food at the retail level.

A key question follows from this analysis: how does UK food inflation compare with that in other countries? Is the UK different or do food prices in fact follow a similar trend to other countries? To that end, Figure 6 charts the CPI for food for a number of OECD countries, some within the EU (Eurozone members and non-members) as well as outside the EU.

Figure 6: Food Price Inflation (CPI) of Selected OECD Countries 1960-2009

Source: OECD

Strikingly, UK inflation appears for the most part to be very similar to most other countries although, with the exception of Japan, the mid-1970s stands out as a period of relatively high food price inflation in the UK. All countries appear to demonstrate a fall in food price inflation over the 1990s and 2000s although as we near the end of the decade food price inflation on the UK appears to accelerate at a greater rate than in the other countries. This suggests that in some ways the UK was hit more significantly by the global food crisis than other countries but this would then, in turn, suggest that it is local or domestic factors that drive UK food price inflation rather than global forces. Section Two examines what these drivers might have been in theory and what therefore could be examined empirically to understand the forces shaping UK food price inflation.

Section 2: The Drivers of UK Food Price Inflation

2.1 Introduction

A distinct and recently growing strand within the academic literature is one that focuses on the movement in, and shocks to, global commodity prices and the subsequent impact on domestic economies, an area that has seen renewed research effort following the commodity price ‘spike’ of 2007/2008. While there are some parallels with the commodity crisis of 1972-4 in terms of the underlying causes of the dramatic increases in world commodity prices, there are also specific differences between the two ‘spike’ episodes. Figure 7 places recent developments in commodity markets in an historical context and shows a series of world (non-oil) commodity prices and oil prices (both in real terms, deflated by the US producer price index) from 1960 through to mid-2009. The figure shows that, though the recent price spikes of 2007/2008 were substantive relative to the level of real prices over the 1990s and 2000s, the spike was nevertheless much less significant than the price changes that occurred in the early 1970s and, with respect to oil, the oil price shock of 1979/1980.

Figure 7: World Real Prices (Monthly, 2005 = 100)

Source: IMF Financial Statistics

Following the commodity crisis of the early 1970s (and the subsequent increase in world oil prices in 1979/1980), there was considerable research into the links between commodity prices and the macroeconomy. To some extent, the economic environment of the 1970s and 1980s was different with commodity price shocks seen to lead or contribute to stagflation while, in the context of recent developments, the recent commodity price has occurred in a period of generally low inflation. Indeed, as we note below, one of the recent strands of current academic research addresses why the links between world commodity markets and the macroeconomy (typically focussed on events in world oil markets) are now weaker, at least as far as developed economies are concerned. In this review, we highlight the recent insights from research into the causes of the commodity price spike of 2007/2008 and the potential impact this has on the macroeconomy and, in particular, inflation. This will serve to inform the econometric strategy underlying the forecasting model to address food price inflation in the UK. The review is divided into three parts. The first focuses on the likely drivers of the 2007/2008 commodity price spike. The second addresses the links between world commodity markets and how world prices are reflected in domestic markets. The final part reports on recent research on the links between commodity prices (principally oil) and inflation.