Particularly in crisis times with an increasing fiscal deficit, high inflation rates and a considerable power shortage there is a necessity for wise political decisions to solve the existing problems in the country. Accurate economic and financial statistics is the basis for more evidence-based decision making and therefore for good governance. One objective of official statistics is to improve the information available to key decision and policy makers in government through the improvement in the scope, quality, relevance and timeliness of statistics. However, the quality of statistics is often limited by low response rates of respondents.
This article is dealing with the treatment of non-responding establishments in official surveys of the Federal Bureau of Statistics (FBS) in Pakistan. A Commission for the Establishment of Business Register in Pakistan has been formed by the Prime Minister, but still surveys in Pakistan suffer from incomplete coverage.1 Up to now Pakistan has no complete business register where all establishments taking part in business activities have to register mandatorily.
The FBS is conducting several surveys like the Census Manufacturing Industries (CMI), the Survey on Small and Household Manufacturing Industries and several other special industry surveys. The CMI for example includes large scale manufacturers with more than ten employees. These establishments are obliged to register their business at the provincial Labor Departments. But not every establishment fulfils this requirement; on the other hand there are 2364 establishments still included in the register, which have ceased their work already. Till now there are more than 13,146 establishments on the mailing list. The aim is to get more accurate figures of the real, up to date situation.
The problems are those establishments, which refuse to answer. The latest CMI shows an increase of participating establishments in this census from 4809 to 6900 firms from 2000/2001 to 2005/2006. But still there is a high rate of non-response units compared with other countries around the world. Overall about 53 per cent of all establishments answered their questionnaires at all. Roundabout only 46 per cent of all establishments qualified for tabulation due to answer their questionnaires fully and correctly (especially regarding input factors). Besides this low share of good respondents, these 46 per cent of all large scale manufacturers provide enough data for an extrapolation due to their weight.
The production of these establishments is estimated with the following method: Legal entities like corporations have to register themselves with balance sheet and profit/loss account in the trade register at Securities and Exchange Commission of Pakistan (SECP). These SECP data offer information on the number of employees as well as the capacities of the (prevailing big) establishments. The total number of establishments covered in the latest CMI was 6417 and total employed persons were about 941,283. The total contribution to gross domestic product (GDP) at basic prices stands at 844 billion PKR. There were 3213 defaulter establishments. With this information the ratio of GDP per employee can be calculated and used for an estimation of non-reporting establishments. The 6417 reporting units were computed activity wise and applied to defaulting units in CMI 2005-06 in order to compare their contribution to GDP. As a result, the contribution to GDP of the large-scale manufacturing sector adjusted for non-response has to be increased for about 18 per cent. As an alternative method, the capacity instead of the number of employees could be used for this estimation. It has to be clear that this extrapolation is only used as help construct due to bad response rates.
One reason for non-response is taxation fear (or even tax evasion and tax fraud). However this is not justified as the statistical law determines that these data can not be used by tax authorities. But according to the Pakistan statistic law, establishments are obliged to answer these kinds of statistical questionnaires. Another argument is the additional workload for the establishments, the so called statistical burden. But therefore the questionnaires may not be the reason, because their size decreased over the years from 28, to 20 to 8 pages in the latest CMI version. But nevertheless the rate of non-response is high. Another reason could be the low probability of being sanctioned for non-response. Therefore, establishments have strong incentives to avoid additional transaction costs of the statistical burden. For that reason, the fine for non-response multiplied with the probability of being caught has to exceed the statistical burden.
The need for a business register can be explained by the following arguments: The frame for (almost) every list-based enterprise survey should be obtained from a (statistical) business register, which ensures a uniform application of concepts (units and classifications) across different surveys. A single business register enables the coordination and data integration across different surveys. Due to accurate calculation of the sample sizes, it allows making use of smaller sample surveys, which decrease the statistical burden, instead of a census including all establishments. And finally, a business register decreases the statistical burden, because it allows a more efficient use of resources than maintaining a separate frame for each survey.
The aim should be the collection of coherent and consistent business data using surveys sampled from high quality business register as far as possible. At the beginning it is very costly for the statistical office to construct a business register solely by survey or observation or secondary sources. Therefore the aim should be a continuously framework of business register, not only a one time exercise for a single year. The leading idea of a business register is that its update is mainly fed by secondary sources, viz. by data which are already available somewhere in the country.
In Pakistan business registration reforms were designed and approved after a consultative process that involved circulating and discussing draft rules with various chambers of commerce, industry, professional bodies, and the public.