IIP

The Index of Industrial Production, something I absolutely detest for the massive revisions and complete WTF in comparison with other economic indicators. But this is the obligatory post on the graphs if only to show you how wild this "index" is.

May 2011 elements came in at 5.6% for the overall index (growth over last year), with mining at 1.4%, Manufacturing at 5.6% and Electricity at 10.29%.

(Click image for a larger graph)

Revisions: the April data, which initially came in at 6.34% has been revised to 5.77%. Feb IIP was increased marginally upwards from 6.48% to 6.67%.

(Note: the initial release for Feb was actually 3.56%, but they have changed the base year from 93-94 to 2004-05, and changed element weights. The new series shows more growth)

When we look at the Use based indexes and growth:

(Click image for larger graph)

If we were to believe this data, then there's a slowdown of sorts in Intermediate goods, consumer durables, and capital goods. The rest are benign.

But you can't trust this because they'll revise the May figures. Twice. Once next month and once in October. When we look at the same data then you might notice that the intermediate goods piece was really strong, something like 10% or more. Our statistical department SUCKS.

Having said all of that, the revisions this time were probably the least botherable. I wonder if data collection is this bad for initial figures or is it just picking numbers out of random body part? Still, it's all we have. I will look at Ajay Shah's seasonally adjusted data to be refreshed to look deepr.

Big IIP move today, coming in at 7.33%. Again, highly unreliable data so it's not something you should take action on. The past revisions have also been minor (just 0.2 to 0.3 on the index levels) which is not newsworthy.

Mining came in at 0.21%, the lowest since Feb 2009. Manufacturing was strong at 7.90%, the highest since October 2010 (which was Diwali, btw). Manufacturing is around 80% of the index, so it determines IIP for the most part.

Electricity, the third player that completes the space, is up 7.19%. Electricity is about 10% of the index.

For the whole year, the growth came in at 7.8%. This is highly strange for a country that has, in nominal terms, grown 20%. I'm not sure if I should even trust this data enough to jump to conclusions, at least not for trading!

Would you like to see a better mapping of this data to see where growth was strong and such? I could create that as a treemap or that kind of visualization.

India's new IIP data was released recently and here you are with the charts:

As you can see, the problem is mining: the growth is just 0.59%. There is a sustained downtrend for two months now, so we might be able to assume the beginning of a problem (IIP data is so volatile and so unreliable that you never ever make a firm decision)

Manufacturing at +3.54% is pretty much par for the course. Remember that February has a budget at the end, and that would hold people's production plans (which is why Feb always has a dip)

Electricity is actually quite strong at +6.75%, but lower than Jan's +10.47%. Power production is strong it seems, though not much of that seems to be coming the way of Gurgaon, going by the number of times the generators need to kick in.

IIP figures seemed to have spooked markets - but everything here is bound to be reviewed and changed. This time, the October 2010 and Jan 2011 data has been changed, but in no major way - Jan has gone up marginally from the earlier reported 3.69% to 3.95%. It's very strange though, that PMI figures vary hugely from the IIP figure (though they should be offset a little bit in terms of time, PMI has shown no signs of contraction)

Look at the equivalent Feb PMI:

I could build complex analysis from the IIP data, but honestly, I do not trust it. Still, it's one of the few data points collected.

The Index of Industrial Production showed a 1.55% growth over last year.

As usual, there were crazy revisions. November data which disappointed everyone with 2.7%, is now revised up to 3.62%. September is up to 4.9% from 4.4%. So this could just be a head fake, for all we know.

And the individual components:

Biggest loser was manufacturing at 1.02%. Weirdly, this makes sense - the Manufacturing PMI dropped in Dec as well, and stayed kinda-sorta flat in Jan.

IIP though is a very shady index to follow, so I would be wary of trading it up on down. A better bet would be the budget.

I don’t trust the (first estimate of) Index of Industrial Production that is released monthly by MOSPI, because it is subject to so many revisions that everything gets invalidated later. Today, they’ve released the index for November and it is at 317.9, a level that points to just 2.7% growth over November 2009.

Immediately after this release markets dropped, but honestly, with a data point that is so volatile, and goes through so many revisions, there’s no point reacting to a headline figure. November’s usually weak, but this November’s the weakest in five years (caveat: could be revised later!)

You might find the sub-item charts more interesting, though.

The Index of Industrial Production (IIP) has grown 5.6% to 309.1 from the 292.8 figure last year.

It is useful to remember that IIP is revised in what can only be described as an obscene manner. Consider:

• The July 2010 figure was revised up from 330.8 to 335 (and another revision will happen later). IIP growth has gone up from the 13% reported to over 15% in July.
• June’s figure was scaled down last month from the originally reported 312.4 to 308.4, taking IIP to 5.76% growth from the originally reported 7.13%

Such 1% revisions are HUGE and therefore current data means nothing. This also means there’s little to infer from even seasonally adjusted IIP figures – the rate of growth is shown from cycle.in:

(This does not include the latest IIP report for August)

Markets supposedly fell on the data release but honestly, a 0.73% fall is just noise, hardly attributable to some specific news.

Earlier posts on IIP:

• June 2010 IIP at 7.13%
• IIP Numbers: Slowdown or “We Don’t Know Yet”?

MOSPI released the Index of Industrial Production (Quick Estimate) for June 2010. The headline general number came in at 7.13% increase over last year. This is the lowest figure since May 2009 and the first sub 10% growth number since Oct 2009.

The revisions were minor – May was revised from the earlier number (312.6) to 312.1. March was revised upwards, from 348.5 to 350.4. Is the best behind us?

Mining was up 9.5% from last year, Manufacturing 7.3% and Electricity 3.5%. Consumer goods grew 8.3%.

Nothing great about this report, as it’s a dull month in general. Diwali based production should perk up the numbers in subsequent months, otherwise we’re heading for the low single digit numbers for a while.

At cycle.in, they show the IIP appropriately adjusted for seasonality. This is majorly negative (month-on-month change, 3 month moving average) – and it only has data till May 2010, June figures yet to be plotted. Let’s wait for that, perhaps next week.

(This is an article I’ve written for a new online site - link later, not something I own)

As we get increasingly connected, it’s easy for information to flow quickly through the ether to the entire world. In a game of Chinese whispers, what is first said varies significantly from how the last person hears it – everyone in the middle adds a little bit of misinformation, however involuntary, and that can change the entire message. Today we have the power of the internet, but with the messages getting increasingly complex, it’s just a sophisticated game of Chinese whispers.

The Ministry of Statistics released the Index of Industrial production (IIP) lately. The headline figure was that we scaled down dramatically from the 16% in April to only 11% in May, something that caused Indian markets to dip in surprise before recovering ground.

But there’s more to it. Looking closer, the IIP figures vary wildly on a month by month basis, due perhaps to the seasonality of such data. Indeed a better view of IIP may be to use seasonal adjustments to make the data comparable month-on-month. Indeed the kind folks at cycle.in have create a seasonally adjusted data series of key indicators, including IIP. With the seasonal adjustment, the IIP change is already below 10% annualized change, for the last two months, not including the current data point (which will likely be added soon).

Last year, the IIP Index shot up from 269 in April to 280 in May – a 3.5% increase which has no parallel increase in this year. This higher base will contribute to a low headline number next month as well (June 2009 saw another 3.5% increase from May 2009), so I would not be surprised to see a figure of 9% IIP growth next month as reported.

Apart from seasonality note that last year was special – there were elections. The seemingly better results –a mandate of more stable majority, and the cutting off of the communist parties – would have prompted industry to rejoice and therefore bump up industrial activity. In a global context, the world has been full of money with governments in the US and Europe attempting to stem falling economic actively by pumping money into banks. A small portion of that money, significant in the Indian context, had seen money flow into emerging economies and contributed to higher industrial activity here.

Finally, the IIP numbers get revised twice, and often, dramatically. The headline figure reported last month – for April 2010 – was over 17%. With the first revision, April has already been reduced to 16% and there will be yet another revision of the April number in August. In the past we have seen changes in this reported figure, of even 3%! (Unfortunately, none of the announcements have been archived at the ministry)

The problem then, of relying on data to make decisions is to attempt to “clean” that data as much as possible. Revision of numbers will never go away, and neither will one-time events that change our economy. We have to learn to interpret data appropriately, instead of calling a slowdown based on a single data point that in itself isn’t reliable.

Increasingly, we rely on analyst “estimates”, and shape our opinions based on whether something was “better than estimates” or worse. This is a format preferred by TV channels and media; an instantly comparable figure makes for a great sound bite. Yet, if you look deeply, analysts have barely ever got it right. Most analysts had predicted a Sensex Earnings Per Share (EPS) of 1,000 for 2009, even in 2010, the EPS is still 840, a good 16% away. In the US, analysts have been consistently wrong for a decade, by as much as 50% when it came to estimating earnings growth of the S&P 500.

It’s not much use to consider analyst estimates as a barometer; we should perhaps rejoice when their estimates aren’t met.

On a happier note, the latest Purchase Managers Index (PMI) for India shows a 2 year high number of 62 for June.

While this tends to be an early indicator, it’s been consistently showing strength; the upcoming result season for listed corporates will confirm. That might just be it right there –confirm a “correction” or “slowdown” with multiple other, independently collected reports, instead of relying on a single, apparently unreliable figure like IIP.