The Purchasing Managers’ Index (PMI) is a widely used economic indicator that gives an insight into business conditions. It is a survey-based diffusion measure provided by Markit, by compiling responses of purchasing managers (executives) of 400 companies every month. The aggregate final measure is a seasonally adjusted number on the scale of 0-100, where values above 50 reflect an increased expectation of overall business conditions compared to the previous month, whereas values below 50 indicate a decline.

The PMI is computed for manufacturing as well as for the services sector. The values are a weighted average of responses related to questions on new orders (30 per cent), output (25 per cent), employment (20 per cent), suppliers’ delivery (15 per cent), and stock of purchases (10 per cent).

The PMI is regarded as a proxy for business conditions as well as a leading indicator of the Index of Industrial Production (IIP). While the IIP numbers are revised, a consensus among economists is that the IIP and PMI plausibly capture a similar trend. This gives us a reason to empirically examine if there is a sustained correlation between the IIP and PMI of the manufacturing sector.

Mapping the correlation

Since the PMI is considered a leading indicator, we carry out two sets of analyses. First, we examine the simple trend by juxtaposing annual growth of the IIP index (manufacturing) and the previous month’s PMI (same results hold with contemporaneous trend analysis). Second, we examine their correlations during the period March 2014 to February 2020 (latest) over rolling windows.

Prima facie , Chart 1 indicates that there are some instances when these indices move in the same direction, but on several occasions, their directions are quite opposite. For example, in the first half of 2015, and since September 2018, PMI and annual IIP growth show little similarity.

To analyse how the correlation between these data series evolves, we plot a four-month rolling window correlation between them over the period. A four-month rolling window correlation considers the correlation among the two indicators over the last four months only, and progressively tracks the evolution with time. As displayed in Chart 2, the relationship between the indices is not consistently positive with time, and there are several instances on a continuous basis where the degree of correlation is strongly negative.

One issue that may be driving this fluctuation in correlation is that the PMI captures the expansion/contraction in the overall business with respect to the previous month, and its correlation with yearly growth on the IIP may lead to erroneous inferences.

To address this concern, we also plot the rolling window correlation between the monthly growth of IIP index and the monthly PMI index over time. We observe a similar pattern of the degree of correlation — fluctuating between positive and strong negative values. Overall, we find that the IIP and PMI indices do not have a sustained positive relationship, as assumed by many practitioners and policymakers. The correlation between the values becomes strongly negative in several instances. The pattern remains similar when we use a 12-month window to estimate the rolling correlation.

Reliance on PMI

This leads us to a natural question, how reliable is the PMI index as a proxy for business conditions of the manufacturing sector in India? Empirical evidence indeed suggests that the PMI is not a consistent leading indicator for industrial production.

If that is the case, reliance on such indicator for macro policy-making is fraught with serious risks. For instance, one can get an impression about how the PMI information is used in policy-making from the discourse of the Monetary Policy Committee (MPC) of RBI. In this regard, we carry out a simple text analysis and measure the number of instances where “PMI” appears in the minutes of the meeting of MPC. We plot the same in Chart 3.

In 2017, the word “PMI” appeared on an average of 3.83 times in every meeting. In 2018, the average frequency increased to 6.5; and 2019 onwards, “PMI” has been mentioned on an average 5.57 times in every meeting. If the mention of “PMI” during the meeting is indicative of reliance of the MPC members on the PMI index for framing monetary policy, then it appears the MPC, inter alia , has been increasingly relying on the PMI. It is not surprising that the RBI is off the mark in forecasting the GDP growth for many quarters.

Lack of clarity?

One can, of course, speculate on why the PMI can be a poor indicator of economic activity in India. As per the official website, the PMI is constructed using the actual changes in the volume of business output. Hence, the quality of assessment of the respondents, and of the ground reality, will drive the observed pattern of changing correlation between the PMI and IIP.

Another possible reason could be that 400 may not be a large enough representative sample to capture the diverse production scenarios of all the firms operating in the manufacturing sector in India. Also, it is not clear exactly how many executives respond every month, and if there is a panel.

Finally, research on the effectiveness of the PMI tracking manufacturing activity in India within and outside the central bank has been quite inadequate. Therefore, as long as it gets wide publicity in the press and policy discourse, the PMI will continue to influence suboptimal policy outcomes. But one wonders if the PMI actually leads or misleads economic activity in India.

Bansal is researcher and Das is faculty at the Misra Centre of Financial Markets and Economy, IIM-A. Views are personal

comment COMMENT NOW