## Charles Bean: Business surveys and monetary policy
Speech by Mr Charles Bean, Deputy Governor (Monetary Policy) of the Bank of England, at the CBI Industrial Trends Survey 50th Anniversary Reception, Centre Point, London, 24 July 2008.
* * *
Good evening! It is a great pleasure to be asked to say a few words to mark the 50th birthday of the CBI's Industrial Trends survey. The birthday boy and I have, in fact, been acquainted with each other for over thirty years - from when I started my professional career in 1975 as a forecaster in the Treasury. In those days, the quarterly Industrial Trends survey was really the only game in town as far as business surveys go. Since then, the field has become more crowded, but the Industrial Trends survey remains a key source of information about business developments.
To celebrate the survey's birthday, I thought that I might say a few words tonight about the part played by business surveys, and intelligence about business conditions more generally, in our setting of Bank Rate each month. Because our interest rate decisions affect UK economic activity and inflation with a rather long, and sometimes rather variable, lag, there is nothing much that we can do to affect the current and near-term rate of inflation. That is already largely baked into the cake. Instead we have to focus on inflation prospects in the medium term - that is, two years or so ahead. So we therefore need to form a judgement on where the economy is likely to be heading. But to do that, we need to start by figuring out where we are now.
The great Danish physicist Niels Bohr is supposed to have observed that 'Prediction is very difficult, especially about the future.' But I don't think he realised just how tough economic forecasting is. Not only is the MPC unsure of the prospects for commodity prices, how the credit crunch will play out and a myriad of other future events, but we don't even have a precise picture of the current state of the economy. As a result, controlling the economy has been likened to driving along a winding road looking only in the rear-view mirror. Indeed, it's worse than that as we don't even have that good a view of where it has been in the past either! It is as if the rear window is also a bit misted up.
Our primary signposts are, of course, the official data provided by the Office for National Statistics for the key macroeconomic variables, such as output, spending, employment, prices, wages and so on. However, we do not rely on the official data exclusively, for a number of reasons. First, it takes time for the official data to be collated and published. For instance, the preliminary estimate of GDP growth in Q2 only appears tomorrow, almost a month after the quarter ended.
Second, early official estimates are often based on incomplete samples, with missing information extrapolated from other indicators. As more information accrues over time, so the data gradually then get revised. And significant revisions can still occur for some years after the event. For an example, look at Chart 1. This plots the initial and current official estimates of four-quarter UK GDP growth, as well as the range of the intervening estimates. As you can see, the picture provided by the mature data sometimes looks quite different to that furnished by the first estimates. For instance, the initial estimates suggested a sharp slowdown in 1998 in the wake of the Asia and Long-Term Capital Management crises, whereas the current estimates suggest that the pace of expansion barely eased. For more recent quarters, the initial and current estimates are closer together, but that may simply reflect the comparative youth of the latest estimates, as well as the fact that the 2007 Blue Book did not involve a full balancing of the National Accounts.
BIS Review 95/2008
Finally, even mature data that has gone through the full revision process may still not give a fully accurate picture because of the intrinsic difficulty of measuring a particular economic concept, for instance the value added by the financial sector. In such cases, the ONS are instead forced to employ proxies, but sometimes these are just not all that good a guide.
The need to form as up-to-date and reliable a picture of the current state of the economy as possible means that the MPC makes extensive use of business surveys alongside the official data. That includes not only the quarterly Industrial Trends survey and its monthly sibling, but also the other, more recent, CBI surveys, such as those covering the distributive trades and service sectors. The various surveys carried out by other organisations, such as the Chartered Institute of Purchasing Supply and the British Chambers of Commerce, are also valuable sources of information for us. Moreover, we have our own network of 12 regional agencies, whose role is to collect intelligence as to what is happening on the ground. Collectively, our regional Agents have about 8000 business contacts and they represent an important additional source of timely information for us.
The uses of the survey data are varied. First, different indicators more often than not seem to tell different stories and we need to figure out which is the right one. For instance, at the current juncture, there is a marked discrepancy between the official data on the volume of retail spending, which points to considerable robustness this year - including whopping growth of 3.6% in May alone, though that was fully reversed in today's release for June - and the Distributive Trades survey, which suggests significant weakness (see Chart 2). While the ONS have suggested that the official data are likely to be better at picking up small retailers and unconventional shopping channels, such as internet purchases, the conformity of the picture given by the Distributive Trades survey with other potential indicators of spending, such as consumer confidence and the Agents reports, have led us to place less weight on the official data on retail spending for the present.
The surveys will not always provide the best coincident guide, however. For instance, during the 1999-2002 period, the official data suggested greater swings in manufacturing output growth than did the Industrial Trends survey (see Chart 3). Now, during this period, we know that there was a rapid expansion in ICT-related activities, followed by a sharp contraction. The simple balance of 'ups' over 'downs' in the Industrial Trends survey could therefore be expected to have failed to reflect adequately the sharp movements in that sub-sector's activity. So, for this period, the official data probably gave a better picture of what was going on.
A second use of surveys is in the application of forward-looking questions to form a view of the near-term outlook. In that regard, questions on business and consumer confidence, and on the trend in orders and backlogs, can give some indication of likely trends in output and spending. The surveys are also useful in forming an assessment of the prospects for capital spending, as this is an area where statistical models are particularly poor - and I should know, having spent some of my formative years at the Treasury unsuccessfully trying to build econometric models to predict investment! Both the CBI and BCC surveys collect information on investment intentions, and it is also something we ask our Agents to take a view on, based on their contact interviews (see Chart 4). We look closely at this information when assessing the outlook for business investment.
Other forward-looking questions can also prove valuable. For instance, at the current juncture, with CPI inflation well above our 2% target, we need to judge how much further it will rise and for how long that might persist. Past experience gives us some guide as to the speed at which higher commodity prices pass down the supply chain and the extent to which those movements are amplified or attenuated by movements in other costs and prices. But the questions on pricing intentions contained in the business surveys represent a valuable additional source of information on near-term inflation pressures.
A third use of surveys is in helping the MPC diagnose the causes of recent developments. Here, the quarterly Industrial Trends survey stands out as particularly useful because it has a rich set of questions that explore companies concerns. For example, it has questions about the factors limiting investment, exports, and output. Responses to those sorts of questions can, for instance, help us work out how businesses have been affected by the recent depreciation of sterling, and the tightening in credit conditions.
A fourth use of the surveys is in providing measures of variables for which there is no official counterpart. In particular, the amount of economic slack in the economy is a key determinant of inflationary pressures. We have official indicators of slack in the labour market unemployment, vacancies and the like - which is one factor determining pay growth. But we have no official measures of the amount of slack within businesses, which potentially affects pricing pressures. Business surveys such as the Industrial Trends survey do, however, ask how intensively firms are working their capital (see Chart 5). And the MPC draws on this
information in forming a judgement about the overall margin of spare resources in the economy.
A fifth, and perhaps less obvious, use of survey information is in refining our picture of the past. I noted earlier that the official data tends to get revised for some while after it is first released. But it turns out that there are some systematic patterns in the historical revision process. Moreover, for some series, we have found a tendency for revisions to be in the direction of reducing the discrepancy with the corresponding survey information. So rather than just taking the latest vintage of official data at face value, we can employ statistical techniques to exploit this information in order to refine our assessment of the past.
An example of this approach, drawn from our May Inflation Report , is shown in Chart 6. This shows our assessment of the range of possible final outturns for four-quarter GDP growth rates for the past six years. The current vintage of official GDP growth data is shown by the green line. The centre of the darkest purple band of the fan reflects our view of the most likely path once the revisions process is complete - a 'backcast', if you like. The fan around this path indicates the uncertainty around this projection, with a 9 in 10 chance of the eventual outturn lying somewhere inside it. The width of this fan is based on the historical experience of data revisions. The fan becomes progressively narrower the further back in time one goes, reflecting the reduced incidence of revisions for more distant periods. For the recent past, the most likely path for the mature estimates of growth lies above the current vintage of data, reflecting our judgement that the business surveys and the past history of revisions together make it somewhat more likely that the official estimates for this period will in future be revised up, rather than down.
I hope that this gives you a sense of the use we make of business surveys, and business intelligence more generally, in making our regular assessment of the state of the economy each month, before taking our decision on the appropriate level of Bank Rate. The timeliness of this information is particularly important at the current juncture, when the economic climate is changing rapidly. The global and UK economies have been subject to two large shocks of uncertain magnitude, duration and impact. The first is the continuing de-leveraging in the banking system and the attendant squeeze on the terms and availability of credit. The second is the fairly relentless rise in commodity prices, especially that of oil, driven in part by the robust growth in demand in the emerging economies. These two shocks both depress output growth, though they have divergent effects on inflation.
In our May Inflation Report , our central expectation was for four-quarter GDP growth to slow markedly through this year, reflecting both subdued real income growth and tighter credit supply (see Chart 7, which also includes the same backcast information as Chart 6). It was then expected to gradually recover through 2009 as credit conditions slowly improved. The corresponding projection for CPI inflation is shown in Chart 8 (note that no uncertainty is shown for the past, as the data for CPI inflation are rarely revised). Our central projection was for inflation to continue to rise during the second half of this year to a peak close to 4% and then - assuming that oil prices stabilised - to fall back towards the 2% target as the growing margin of spare capacity in the economy held back pay and price increases. Since May, inflation has turned out higher than expected and oil prices have risen further, notwithstanding some moderation in the past week or so. But the news on activity has not been very good either, with some distinctly downbeat business surveys, including yesterday's Industrial Trends survey, further weakening in the housing market and continued fragility in the financial sector. All this points to a worse near-term prospect for both growth and inflation. The net impact of this on the outlook for inflation in the medium term is less clear, though for our considered view on that, you will have to wait until our August Inflation Report is published in a couple of weeks time.
Moreover, there are considerable risks surrounding the central outlook. A particular risk on the upside is that the extended period of elevated inflation leads to an acceleration in pay and causes inflation expectations to rise. That could lead the higher inflation to become embedded. But on the downside, there is a risk that the credit crunch leads to a deeper and more prolonged slowdown, leading to greater slack in the economy and causing inflation to undershoot our target.
In setting policy, we have been trying to balance off these risks against each other. In making that judgement, the more accurate a picture we have of the economy, the better. So the Industrial Trends survey, as well as the other business surveys, will continue to be a crucial
input into forming that picture. Let me therefore conclude by wishing Many Happy Returns to the Industrial Trends survey on the occasion of its 50th birthday. We hope that you and your younger siblings will continue to flourish in the years ahead. Thank you!