Manufacturing Slowdown in India

Nominal value added is the difference between the value of an industry’s output and the cost of raw materials or intermediate inputs. Real value added is obtained by deflating nominal value added by suitably chosen price indexes. A proper estimate of real value added is necessary for various reasons such as maintaining the national income accounting identities in real as well as nominal terms, thereby ensuring the equality of the gross domestic product (GDP) when measured by value added, income and expenditure approaches (Sato 1976), as well as performing productivity analyses of an industry by separating out the contribution of primary inputs from economies of scale and technical change, among others (Cassing 1996).

The two basic approaches to deflating nominal value added are the single deflation approach and the double deflation approach. The single deflation approach deflates nominal value added by an output price index, while the double deflation approach deflates outputs and material inputs separately by their respective price indexes. The Central Statistics Office (CSO) uses the single deflation approach to measure manufacturing real valued added in India (CSO 2015). However, for the sake of consistency between GDP figures from the value added and expenditure approaches, the United Nations System of National Accounts (SNA) recommends the use of the double deflation approach to create a Laspeyres-type index of manufacturing real value added (UN 2008). Implementing the latter in the Indian context is complicated by the absence of an official intermediate inputs price index.

Laspeyres-type double deflation indexes of real value added mitigate the unwanted effects of changes in relative prices of material inputs to outputs (terms of trade) on real value added, by deflating material inputs and outputs by their respective price indexes (Hansen 1974). These terms of trade effects are unwanted since they would otherwise get conflated with the effects of changes in physical inputs and outputs, thereby polluting the measurement of physical productivity of primary inputs, one of the main aims of correct measurement of real value added (Sato 1976). The extent of terms of trade bias in a single deflation real value added measure is therefore an important empirical issue, which is investigated in this article.

We measure real value added in the Indian manufacturing sector for the period 2011–12 to 2016–17 using the double deflation approach. We find that the official figures understate manufacturing real value added during the period 2011–12 to 2013–14, and overstate it thereafter, as well as miss an apparent manufacturing contraction that occurred in 2014–15. Our results are corroborated by the movement of high frequency indicators that are correlated with manufacturing activity.

Double Deflation

Balakrishnan and Pushpangadan (1994) measured manufacturing real value added in India using the double deflation approach, as a prelude to measuring total factor productivity growth in manufacturing during the decades of the 1970s and 1980s. The input price deflator was a weighted index of wholesale prices of major input groups, with weights calculated from the 1973–74 input–output transactions table of the CSO. Inputs were sorted into 19 groups according to the availability of wholesale prices that most closely represented them.

Balakrishnan and Pushpangadan (1994) found that the relative prices of inputs rose during the 1970s, and then declined during the 1980s, while value added under double deflation was higher than that under single deflation for most of the 1970s and 1980s, with the gap between the two reaching 52.6% by the end of the period. The results do not vary much when they use weights from the 1983–84 input–output transactions table as a robustness check. Since the work of Balakrishnan and Pushpangadan (1994), other attempts have been made at generating double deflation value added figures for manufacturing, briefly reviewed in Balakrishnan and Pushpangadan (2002).

In more recent work, Rajakumar and Shetty (2015) generate a manufacturing real value added series for India for 2011–12 to 2013–14 using the double deflation approach. They construct an intermediate input price index using input–output tables for 2007–08 as well as data from the 2004–05 series of the wholesale price index (WPI). The shares of various commodity groups in manufacturing sector’s consumption of intermediate inputs from the input flow (absorption) matrix were used as weights for the corresponding commodity groups in the WPI to generate an intermediate input price index as a weighted average of WPI of the corresponding commodity groups. The WPI of individual commodity groups in the 2004–05 series were indexed to 2011–12 using the splicing method. The output price index used was the implicit deflator from the 2011–12 series of manufacturing gross value added (GVA). Rajakumar and Shetty (2015) found that the size as well as growth rate of manufacturing real value added under double deflation is lower in 2012–13 and 2013–14 than that reported by the CSO (which follows the single deflation method).

Their findings were criticised by Dholakia (2015) on the grounds that the intermediate input price index constru­cted by Rajakumar and Shetty (2015) did not take into account the price of construction and services inputs into the manufacturing sector, although the two accounted for 15.4% and 17.5% of the total inputs respectively. To the extent that construction and services input prices move differently from those of commodity inputs, the intermediate input price index of Rajakumar and Shetty (2015) will be biased. Using the GVA deflator (ratio of value added at current and constant prices) for construction and services as a proxy for their price levels, Dholakia finds that construction and services inflation rates were higher than that of commodity inputs. This implies a negative bias in the intermediate input price index of Rajakumar and Shetty (2015).

A negative bias in the intermediate input price index caused by faster growth in the prices of the omitted inputs, that is, construction and services will translate into a negative bias in the double deflation value added. When taking construction and services inputs into account and deflating them by their associated GVA deflators, Dholakia (2015) therefore finds that manufacturing double deflation value added is greater than that of Rajakumar and Shetty (2015) in levels as well as growth rates.

We extend the work of Dholakia (2015) in two ways. First, we make use of more recent data to generate a longer manufacturing real value added series, spanning 2011–12 to 2016–17. Second, we use the newly created producer price index (PPI) to deflate services inputs into the manufacturing sector, rather than the services GVA deflator.

Data and Methodology

We create and compare two indexes of real value added for India: (1) manufacturing single deflation (MVASD), and (2) manufacturing double deflation (MVADD). The formulas for single and double deflation measures for real value added are displayed in Equations (1a) and (1b),

where SL refers to a single deflation measure of real value added index, DL refers to a double deflation measure of real value added index, t refers to the current period, 0 refers to the base period (2011–12 for all indexes), P is a vector of gross output prices, Q is a vector of gross output levels, W is a vector of intermediate input prices, and X is a vector of intermediate input levels. A real value added index for period t is the ratio of real value added at period t to real value added in period 0. PtQt is gross output at current prices, P0Qt is gross output at constant prices, WtXt is intermediate inputs at current prices, W0Xt is intermediate inputs at constant prices, and POt is an output price index (which is just the GVA deflator in the single deflation approach).

Data on PtQt, P0Qt, Wt Xt, W0 Xt, and the GVA deflator for the manufacturing sector is available for the period 2011–12 to 2016–17 from the National Accounts Statistics 2018 Statement 1.5 (NAS 2018). Note that SL is constructed from this data, following the single deflation approach. Since SL is constructed using the single deflation approach, it suffers from terms of trade bias since it does not take into account intermediate input prices.

DL is a double deflation real value added index of Laspeyres type. Note that DL will equal SL if the output and intermediate input price indexes coincide exactly. Measuring the terms of trade bias, defined as the difference between DL and SL, is one of the objectives of this paper. NAS 2018 does not provide a separate implicit deflator for intermediate inputs since it adopts the single deflation approach to measuring manufacturing real value added (we therefore cannot use W0Xt from NAS 2018 since it is equal to Wt Xt /POt). We must therefore compute a Paasche price index for intermediate inputs , and use it to deflate the nominal value of intermediate inputs Wt Xt to recover the real value of intermediate inputs W0 Xt, which can then be used to compute DL.

It can be shown that when intermediate input prices are higher than output prices that is POt, we have DLt>SLt and vice versa. The logic for this is as follows. When intermediate input prices are higher than output prices, real value of intermediate inputs is lower under double deflation than under single deflation approach, whereas real value of output is the same under both approaches. Therefore, the difference between real value of output and real value of intermediate input, that is, value added is higher under double deflation approach than under single deflation approach when intermediate input prices are higher than output prices. The opposite is true when intermediate input prices are lower than output prices.

The India KLEMS database provides nominal and real values of the three major intermediate input categories that is energy, materials, and services used in the manufacturing sector for the period 2011–12 to 2016–17, with base year 2011–12. The basic source of data for intermediate input use in manufacturing is the input flow (absorption) matrix for the years 2007–08 and 2013–14, with suitable interpolation to ensure consistency of the intermediate input time series in current prices with the figures from the NAS. Let WtE XtE, WtM XtM and WtS XtS be the nominal values of energy, materials and services inputs into the manufacturing sector at time t. Let W0E XtE, W0M XtM, and W0S XtS be the corresponding real values at time t. Then, the Paasche intermediate input price index for the manufacturing sector is displayed in Equation (2).

Note that both construction and services are included in intermediate inputs in the India KLEMS database, thus addressing the criticism of Dholakia (2015). While commodity inputs at current prices are deflated using the appropriate WPI, services inputs at current prices in the India KLEMS database are deflated using the implicit services GVA deflator from the NAS, which is the same treatment of services as in Dholakia (2015). However, this is not consistent with a double deflation approach, since the same deflator is being used to deflate services outputs at current prices. The input producer price index (PPI) can be used as an alternative to the services implicit GVA deflator to deflate services inputs at current prices. The input PPI measures the prices of goods and services as they enter the production process, that is, purchaser’s prices, and are suitable for use as deflators in National Accounts (GOI 2017).

For the services sector, the input PPI is constructed on the basis of price data from the CPI as well as the business service price index (BSPI) put out by the Office of the Economic Adviser in the Department for Promotion of Industry and Internal Trade, Government of India. Weights for the input PPI are based on the input structure reflected in the Use Table 2011–12. Choice of CPI price data to generate input PPI is justified by common point of purchase and sale of services, as well as a competitive environment ensuring that rates of change of producer and consumer prices remain close to each other (GoI 2017). As opposed to this, the implied GVA deflator for services (ratio of nominal to real value added of services) is derived using a combination of WPI (despite the fact that it does not cover services), CPI, and quantum indexes to deflate nominal value added to recover real value added.

Since we adopt the double deflation approach in this paper, we deflate WtS XtS by the services input PPI to get an alternate measure of W0S XtS used in the manufacturing sector. For the sake of comparison, we also deflate WtS XtS by the implied GVA deflator for services to get an alternate measure of W0S XtS that is analogous to the approach in the Indian KLEMS database as well as Dholakia (2015). Correspondingly, we get three measures of manufacturing real value added index, that is, MVASD, MVADD (gvas), and MVADD (ppis) for single deflation, double deflation using implied GVA deflator for services, and double deflation using services input PPI respectively.

Results and Discussion

The Paasche output and input price indexes for manufacturing are displayed in Figure 1. Clearly, they do not move together, with the input price index (gvas and ppis) exceeding the output price index till 2013–14, and falling below the output price index from 2014–15. Further, the input price index (ppis) is everywhere below the input price index (gvas). We would therefore expect that MVADD (gvas and ppis) will exceed MVASD till 2013–14 (indicating a negative terms of trade bias in MVASD) and thereafter fall below MVASD from 2014–15 (indicating a positive terms of trade bias in MVASD). Further, MVADD (ppis) will lie everywhere below MVADD (gvas). This is exactly what we observe in Figure 2. MVASD thus understates the extent of real value addition in the first half of the period, and subsequently overstates it. The terms of trade bias, defined as the percentage difference between MVASD and MVADD is displayed in Figure 3 (p 63). This bias is quite large, reaching a maximum of 12.98% (gvas) and 26.37% (ppis) in 2015–16.

The MVADD (ppis) figures show a contraction in manufacturing value added in 2014–15. This contraction is not captured by the MVASD data, which instead shows an expansion. However, all four high frequency (monthly) indicators in the Mint Macro Tracker that are correlated with industrial sector performance show signs of a contraction in 2014–15, as reflected by the second order polynomial fitted to the data (see Figures 5–8). These indicators are core sector IIP, bank’s non-food credit, rail freight traffic, and manufacturing purchasing manager’s index (PMI). Data for these indicators were extracted from Kwatra and Bhattacharya (2019). The high frequency indicators support the idea of a manufacturing contraction in 2014–15, as reflected in the MVADD (ppis) figures. Growth rates of MVASD and MVADD are displayed in Figure 4, showing faster growth of MVADD than MVASD till 2013–14, followed by slower growth in 2014–15 and 2015–16.

In conclusion, we find that the double deflation approach to measuring real value added provides significantly different conclusions about the performance of the manufacturing sector both in terms of levels as well as growth rates. These differences are driven by differences in the movement of the corresponding output price indexes and the intermediate input price indexes. The official figures understate manufacturing real value added during the period 2011–12 to 2013–14, and overstate it thereafter, as well as miss an apparent manufacturing contraction that occurred in 2014–15. This contraction is corroborated by the movement of high frequency indicators that are correlated with manufacturing sector performance.

References

Balakrishnan, P and K Pushpangadan (1994): “Total Factor-productivity Growth in Manufacturing Industry: A Fresh Look,” Economic & Political Weekly, Vol 29, No 31, pp 2028–35.

— (2002): “TFPG in Manufacturing: The 80s Revisited,” Economic & Political Weekly, Vol 37, No 4, pp 323–25.

Cassing, S (1996): “Correctly Measuring Real Value Added,” Review of Income and Wealth, Vol 42, No 2, pp 195–206.

CSO (2015): “Changes in Methodology and Data Sources in the New Series of National Accounts: Base Year 2011–12,” Central Statistics Office, New Delhi: Government of India.

Dholakia, R (2015): “Double Deflation Method and Growth of Manufacturing: A Comment,” Economic & Political Weekly, Vol 50, No 41, pp 88–90.

GoI (2017): “Report of the Working Group: Producer Price Index,” Office of the Economic Adviser, Department of Industrial Policy and Promotion, Government of India.

Hansen, B (1974): “A Proposed Real Net Output Index: A Comment,” Review of Economics and Statistics, Vol 56, No 3, pp 415–16.

Kwatra, N and P Bhattacharya (2019): “Economy Is Slowing: Mint’s Macro Tracker Shows,” Livemint, 30 Jannuary, https://www.livemint.com/politics/policy/economy-is-slowing-mint-s-macro-tracker-shows-1548781168732.html.

Rajakumar, J D and S L Shetty (2015): “Gross Value Added: Why Not the Double Deflation Method for Estimation?” Economic & Political Weekly, Vol 50, No 33, pp 78–81.

Sato, K (1976): “The Meaning and Measurement of the Real Value Added Index,” Review of Economics and Statistics, Vol 58, No 4, pp 434–42.

UN (2008): System of National Accounts 2008, New York: United Nations.