Poverty data: Chor machaye shor

Independent analysts applying our methodology to our data to get our results do not validate fudged poverty figures.

The recent debate on poverty focuses on the accuracy or otherwise of the poverty estimate of 2007-08 and the questionable alternative estimates. As a result, most writers have fallen in the trap laid by those who still have to be held accountable for sowing the seeds of the continuing chaos in our statistical system. They have also succeeded in shifting the terms of the debate to methodology of estimation and its validation. These are the fringes, not the kernel. The question is whether the data to which methodologies have to be applied is credible. When the World Bank or poverty veterans  like Professor Kakwani say they have no problem with the methodology, they are not saying anything. Methodologies differ with the tastes of the users. They are endorsed so long as they are applied consistently overtime. This is what is meant by validation - independent analysts applying our methodology to our data to get our results. In the process, they are not validating our data. In fact, they have no way of validating such a large data set. They do not either have access, or it is limited to the final dataset. The process of collection of data and the manner of its processing is kept well beyond them.

Now the data can only be messed up by influencing those who collect and process it. This is exactly what happened. An official poverty line, broadly accepted after some initial debate, was applied for the first time to the dataset relating to 2000-01. Though it understated poverty in a period of drought, the resulting estimate of 32.1 percent was viewed bad optics by the finance minister of the military government. The result was withheld and his economic adviser, a contract employee always too eager to please the boss to keep the job, forced the staff of the federal bureau of statistics to retract, but without success.

Systematic preparations were made for the next survey due for 2004-05. By then the finance minister had been elevated to the office of prime minister. A pliable additional secretary in-charge of statistics division who, instead of appointing the senior most officer in the federal bureau of statistics as director general, assumed the office himself, and allowed all the prime minister’s men to micro manage the 2004-05 survey from the word go. They virtually laid siege to the main computer centre of the federal bureau of statistics. Having ensured a dataset of their liking and clandestinely using the staff of the poverty centre of the planning commission to ensure what might be the result, the data was formally handed over to the planning commission. Any one using this dataset and applying the same methodology would come out with the same result. Calling this validation would require a change in the standard understanding of the word.

As I noted in my last column in these pages, the federal bureau of statistics has washed its hands off the 2004-05 HIES (household integrated economic survey) by ignoring to mention it in its latest report. The report introduces the latest HIES thus: “The income and consumption module [for HIES 2010-11] is exactly the same which has been used previously for the HIES 2001-02, HIES 2005-06 and HIES 2007-08.” It’s the data, stupid! The current debate reminds me of the Urdu saying “Chor machaye shor”.

Published in The Express Tribune, October 28^th, 2011.