A slow quick estimate
An inelegantly constructed index results in big swings in the Index of
Industrial Production, making it not the ideal basis for swift policy
responses.
Should alarm bells have gone off when it surfaced that manufacturing output in India grew at just 5.6 per cent in August after recording 15.2 per cent expansion the preceding month? Or, was it just a blip? For Indian policy makers, fortunately, the situation is not alarming - a plot of industrial output data since 2002 reflects a steady uptrend, more so in recent years.
The Index of Industrial Production, or IIP, data aggregated over April-August 2010 shows factory output grew 10.6 per cent over the same months of 2009 - nearly double the industrial growth in April-August 2009 over 2008.
But the bad news is that the IIP has time and again displayed a high degree of volatility making it difficult for mandarins to ready timely policy responses. Especially for the Reserve Bank of India that watches the IIP closely to see how manufacturing, which makes for 15 per cent of the GDP, is chugging along. As the central bank fretted in its September policy review, "... the high volatility in past two months raises some doubts about how effectively the index reflects the underlying momentum in the industrial sector."
Typically, provisional IIP data is released about six weeks late for each month (the August data was out on October 12) and is updated about four weeks later. Often, the final data shows a marked variation from provisional data. The July IIP final estimate, for instance, was a full two percentage points more than the quick estimate. But that's only a small reason why things get hairy. The bigger problem is that depending on IIP data for quick monetary decisions, well, is not advisable.
The whole idea of having a quick estimate series such as the IIP is to pick up early signals from the ground for remedial policy action. But high volatility forces policy makers to pussyfoot on decisions for a couple of quarters until they can be reasonably confident of the underlying trend, which in turn defeats the purpose of having a quick estimate.
Imagine the RBI having to wait for months for a reliable trend to emerge from IIP data before it can decide on interest rates policy. "The problem with the IIP is that we can't get a good estimate of short horizon estimates," says India's Chief Statistician T.C.A. Anant.
To be sure, that's a problem the government has been struggling with for more than 15 years - the last revision of the index was done in 1993-94 - and Anant is not sure solutions are easy to come by. Several committees and experts have failed to find a substitute for the series or been able to suggest ways to smoothen it out.
According to Anant, the IIP problem does not relate to methods of data gathering but the way the index is structured. "The volatility has been a consistent trend since 2002 and a data problem can't be so systematic," explains Anant.
Primary among the inherent limitations of the IIP is that it captures trends of a fixed basket of products. It does not capture new products in the market or entry and exit of producers. So, say five companies produce a good and a new company comes into this market. This will make the output volatile owing to increased production by the new company but also due to the response of the older ones. Or, there could be variation in demand and output of a product as a result of an ad campaign. But the IIP is structured in a way that it cannot correct for the volatility arising out of such normal market behaviour.
Then, the long gaps in revising the basket of products it monitors does not help IIP's case. So, there are products like typewriters and black and white televisions still tracked by the index. Or, something like mobile phones - at least 100 million are manufactured a year - has not been assigned adequate weightage.
Of the various industries that the IIP data captures, the output of capital goods has contributed maximum volatility over the last couple of months to the IIP. Growth in output of Machinery and Equipment goods decelerated to 0.4 per cent in August 2010 over the same month of the previous year. The growth in July was much higher at 57 per cent over July 2009. A simple explanation for the volatility in this product group is that capital goods makers are likely to be sensitive to some corporate accounting practices and tax payments-related concerns among clients guiding production decisions. So the lumpiness in production of capital goods could be traced to accounting and reporting practices of companies or inventory management.
The trend in a large conglomerate like the aluminium-to-viscose fibre Aditya Birla Group seems to confirm this. "In the Aditya Birla group, capital expenditure is very robust and the outlook very positive. But the expenditure does tend to be lumpy as there are global tenders involved, orders have to be placed and the nature of these goods makes the expenditure chunky," explains Ajit Ranade, Head of the Corporate Economic Cell at the group. But such lumpiness should get averaged out in a well-constructed index, says Anant, owing to various companies exhibiting such trends. The weakness of the IIP is that the averaging in the capital goods category does not smoothen out the series.
Can Anant fix an index that no one has been able to for long years? One idea the country's chief statistician is toying with is alternatives that work as a quick indicator for industrial growth. These could substitute or read in tandem with the IIP. One such quick estimate that he may use as a proxy for industrial growth is a new quick turnaround short term survey that the Labour Ministry is experimenting with for employment projections. He is also examining if data on tax collections disaggregated in a certain way can be used to serve the purpose of making short horizon projections on industrial growth.
But, as he embarks on the exercise, of one thing he can be sure: there is unlikely to be a quick fix to the problem.
But the bad news is that the IIP has time and again displayed a high degree of volatility making it difficult for mandarins to ready timely policy responses. Especially for the Reserve Bank of India that watches the IIP closely to see how manufacturing, which makes for 15 per cent of the GDP, is chugging along. As the central bank fretted in its September policy review, "... the high volatility in past two months raises some doubts about how effectively the index reflects the underlying momentum in the industrial sector."
Typically, provisional IIP data is released about six weeks late for each month (the August data was out on October 12) and is updated about four weeks later. Often, the final data shows a marked variation from provisional data. The July IIP final estimate, for instance, was a full two percentage points more than the quick estimate. But that's only a small reason why things get hairy. The bigger problem is that depending on IIP data for quick monetary decisions, well, is not advisable.
The whole idea of having a quick estimate series such as the IIP is to pick up early signals from the ground for remedial policy action. But high volatility forces policy makers to pussyfoot on decisions for a couple of quarters until they can be reasonably confident of the underlying trend, which in turn defeats the purpose of having a quick estimate.
Imagine the RBI having to wait for months for a reliable trend to emerge from IIP data before it can decide on interest rates policy. "The problem with the IIP is that we can't get a good estimate of short horizon estimates," says India's Chief Statistician T.C.A. Anant.
To be sure, that's a problem the government has been struggling with for more than 15 years - the last revision of the index was done in 1993-94 - and Anant is not sure solutions are easy to come by. Several committees and experts have failed to find a substitute for the series or been able to suggest ways to smoothen it out.
According to Anant, the IIP problem does not relate to methods of data gathering but the way the index is structured. "The volatility has been a consistent trend since 2002 and a data problem can't be so systematic," explains Anant.
Primary among the inherent limitations of the IIP is that it captures trends of a fixed basket of products. It does not capture new products in the market or entry and exit of producers. So, say five companies produce a good and a new company comes into this market. This will make the output volatile owing to increased production by the new company but also due to the response of the older ones. Or, there could be variation in demand and output of a product as a result of an ad campaign. But the IIP is structured in a way that it cannot correct for the volatility arising out of such normal market behaviour.
Then, the long gaps in revising the basket of products it monitors does not help IIP's case. So, there are products like typewriters and black and white televisions still tracked by the index. Or, something like mobile phones - at least 100 million are manufactured a year - has not been assigned adequate weightage.
Of the various industries that the IIP data captures, the output of capital goods has contributed maximum volatility over the last couple of months to the IIP. Growth in output of Machinery and Equipment goods decelerated to 0.4 per cent in August 2010 over the same month of the previous year. The growth in July was much higher at 57 per cent over July 2009. A simple explanation for the volatility in this product group is that capital goods makers are likely to be sensitive to some corporate accounting practices and tax payments-related concerns among clients guiding production decisions. So the lumpiness in production of capital goods could be traced to accounting and reporting practices of companies or inventory management.
The trend in a large conglomerate like the aluminium-to-viscose fibre Aditya Birla Group seems to confirm this. "In the Aditya Birla group, capital expenditure is very robust and the outlook very positive. But the expenditure does tend to be lumpy as there are global tenders involved, orders have to be placed and the nature of these goods makes the expenditure chunky," explains Ajit Ranade, Head of the Corporate Economic Cell at the group. But such lumpiness should get averaged out in a well-constructed index, says Anant, owing to various companies exhibiting such trends. The weakness of the IIP is that the averaging in the capital goods category does not smoothen out the series.
Can Anant fix an index that no one has been able to for long years? One idea the country's chief statistician is toying with is alternatives that work as a quick indicator for industrial growth. These could substitute or read in tandem with the IIP. One such quick estimate that he may use as a proxy for industrial growth is a new quick turnaround short term survey that the Labour Ministry is experimenting with for employment projections. He is also examining if data on tax collections disaggregated in a certain way can be used to serve the purpose of making short horizon projections on industrial growth.
But, as he embarks on the exercise, of one thing he can be sure: there is unlikely to be a quick fix to the problem.
