India’s chief statistician TCA Anant says this in this interview. I would think both data and analysis are connected. Poor data surely leads to poor or distorted analysis. Though, this does not mean good data always leads to good analysis.
He says we need to look beyond monthly volatility and look at compsition of data:
Analysts such as Morgan Stanley’s Ruchir Sharma question the credibility of official data, pointing to a dichotomy between GDP growth numbers and corporate performance. Others say falling industrial output doesn’t add up with rising inflation. What would you say to them?
The Index of Industrial Production or IIP is and has always been a volatile series. To correlate it on a month-on-month basis with inflation is not statistically desirable or advisable. The IIP monthly data faces a number of sources of volatility which include seasonal factors and includes in the context of India, a complicated seasonal, but not regularly predictable set of holidays. For example, Dussehra and Diwali occur every year, but in a band of 30 to 45 days. There are other regionally significant festivals some of whom are very predictable in there timing while others are not.
This complicates the volatility in Indian industrial data, which needs to be appreciated when you work with disaggregated monthly data. Remember this volatility is missing if you work with the annual data. But when you come down to the monthly level, a very large number of commentators are oblivious to this.
What about inflation?
The two major indicators of price movement — consumer price index (CPI) and wholesale price index (WPI) — cover very different baskets of commodities with different weightages. People talk about a discrepancy between CPI and WPI. If looked at carefully on a common set of commodities, the two indicators behave similarly. It’s not that something is being done differently in the WPI, but different commodities are behaving differently. This is a reality – all commodities are not experiencing the same form of price movement today, unlike in the past when there was generalised inflation and every commodity was seeing a rise in prices, albeit at different rates. We are now seeing a situation where some commodities are seeing a fall in prices, while some are seeing a rise. That is captured based on whether they enter the consumption basket or not. This complexity in price movement needs to be understood for its implications on what it says about the economy and what does it say about incentives to different segments of the economy. It is not a problem with the data. It is a problem that the analyst needs to get lot more sophisticated in their analysis.
These are perennial issues. In countries like US which have so many data releases, one is never sure about direction of economy. The data points to different directions most of the time.
But one can rarely question the quality of data which has become a seething problem for any kind of analysis..