FE Edirorial: Year is a long time in economics

Wed, Oct 14 02:14 AM

These columns have argued for changing the way official economic data is processed. For example, we argued that month-on-month, seasonally-adjusted data gives a more realistic picture of current industrial growth—healthy, but not exuberant and, therefore, requiring more support. But month-on-month seasonally-adjusted data has abiding significance in all areas. The following simple example will make this clear. Suppose for good A, the production in month X in 2008 was 100 units and in the same month in 2009 was 110 units. The way government processes data—year-on-year—this will be interpreted as a 10% industrial growth rate. But how has the production of good A been recently? That's what policymakers should want to know. Did the 100 units to 110 units change over 12 months come because production rose quickly, then declined sharply and then stayed steady? Or did production of good A remain steady until, say, a quarter before month X, 2009 and then start moving up steadily? The two scenarios say very different things about recent trends in production. The first one suggests that there was a sharp spike early in the 12-month period, but that recent growth momentum is weak. The second suggests that things are looking up in recent months. These have very different implications—for policymakers, for independent economists and for media analysts. When we focus on production on a year-on-year basis, as we tend to do in India, we miss out on the process described above.

The way this problem has been addressed in developed countries is to, of course, focus on monthly changes. There is, however, one problem with monthly changes. If there is seasonality in the data, it might suggest there is a rise in production and we might start taking policy action, such as reversing the monetary and fiscal stimuli provided to the economy. But that would be a mistake. So, while economists like to focus on monthly production, they do so after seasonally-adjusting the data. Some of the simplest techniques available to do this are nearly 100 years old. The latest techniques are ubiquitous now; software packages to apply them are available for free download on the Internet. A small team of 10 to 15 economists with training in econometrics and computer programming can do this on a regular basis for most of the important macroeconomic data available. Statistical departments in advanced countries release seasonally-adjusted data, in addition to the raw data. These numbers are then picked up by the central bank for policy analysis and by the government for its communication with policymakers. The media in the West uses the same data. There's no earthly reason why official data processing here can't change. So, can we see that change?

Inside News

Editor's picks