ISSN (Print) - 0012-9976 | ISSN (Online) - 2349-8846
-A A +A

Did Public Investment Crowd Out Private Investment in India?

Honey Karun (honeykarun@gmail.com) is with Senior Resident Representative Office, International Monetary Fund, New Delhi, India. Hrishikesh Vinod (vinod@fordham.edu) is with Fordham University, New York, United States. Lekha S Chakraborty (lekha.chakraborty@nipfp.org.in) is with the National Institute of Public Finance and Policy, New Delhi, India.

The maximum entropy bootstrap method is applied to overcome the econometric constraints of using a short time series after the publication of a new macroeconomic series in India. A short time series (quarterly data) of stationary and non-stationary variables between 2011 and 2016 is used to confirm the positive role of public infrastructure investment. The significant result has policy implications in terms of the current debate, whether public investment “crowds in” rather than “crowds out” private corporate investment in India.

Private investment in India has ­averaged around 25% of the gross dome­stic product (GDP) during 2004–05 to 2015–16, wherein both the corporate and household sectors consistently contributed more than 10%. Public sector contributed an average of 8%–8.5% of GDP during the same period.

Successive economic surveys in India (for instance, 2012–13, 2013–14, and 2014–15) have highlighted several factors causing the decline in private invest­ment over time. The Economic Survey 2013–14 stressed the severity of challen­ges in financing private investment. It also argued that high and persistent infla­tion, along with lower real interest rates, are reducing private savings, thus reducing the supply of funds. Accordingly, the survey urged policy measures aimed at reducing the fiscal burden (through fiscal consolidation), stabilising inflation, and reduction in resource pre-emption, thereby allowing more fin­ancial space for private investment (or reduced “crowding out”).

The Central Statistics Office (CSO) of India introduced a new series of national accounts, with certain revisions in the methodology for estimating gross value added (GVA) and GDP, which provides data at 2011–12 prices.1 Thus, a limited number of observations available posed challenges to perform meaningful time series analysis.

This article hence uses the maximum entropy bootstrap method to overcome the econometric constraints of using a short time series after the publication of the new macroeconomic series in India. The results reinforce the crow­ding-in properties of public investment in India.

Literature Review

In the Indian context, Chakraborty (2007, 2016) attempted to explore both real and financial aspects of the crowding-out argument and found no evidence for either. In recent years, Bahal et al (2015) observed crowding-out effects on private investment during 1950–2012, whereas the opposite results were highlighted for the post-1980 period. Dash (2016) found evidence for crowding-out of private investment for the period 1970–2013, which was subdued during the post-liberalisation period and a positive impact of public infrastructure investment on private investment in the short run.

Mallick (2016) attributed the crowding-out effects of public investment during 1970–2013 to non-infrastructure government investment. Chhibber and Kalloor (2016) argued for crowding-in effects of public investment on aggregate and sector-wise (corporate and non-corporate) private investment for 1980–2014.

The empirical literature reviewed here relies on autoregressive distributed lag (ARDL) and vector autoregressive (VAR) models. The ARDL and VAR models often involve differencing or de-trending of variables to deal with the problems associated with the ubiquitous non-stationarity of underlying macro­economic time series. Moreover, these models often yield insignificant results when the time series is short.

In this article, maximum entropy boot­strap (“meboot”) is considered based on Efron (1979) for exploring determinants of private investment in India.

The “meboot” algorithm is a seven-step procedure that allows one to generate replicates or “reincarnations” of the original series, as termed by Vinod (2004), to be used for statistical inferences. The meboot resamples allow overcoming the unit root and structural change pretest problems, while avoiding any differencing-type transformations of original time series simply for ensuring the stationarity assumption.2 In addition, the constructed ensembles have the property of retaining the overall shapes of autocorrelation and partial autocorrelation functions of the original time series data, without imposing parametric constraints.

Figure 1 (p 22) shows the actual data on private investment and a sample of three replicas generated from the meboot algo­rithm. It shows that the basic shape of the non-stationary I(1) series is retai­ned in each replica as the resamples are strongly dependent on it.

Interpreting Data and Model

The determinants of private investment are explored following Chakra­borty (2007, 2016) by incorporating interest rates (both short and long term) in the model equation as below to gauge the impact of interest rates on private corporate investment:3

Ipvt = α + β1Ipub+ β2ir + β3Cpvt+ β4Kforgn+ β5Y*+ et

... (1)

where Ipvt= private investment, Ipub= public investment, ir= real interest rate (using two versions: short- or long-term rate), Cpvt= credit to the private sector, Kforgn= foreign investment capital flows, and = output gap. Both the price and quantity of credit variables are added in the model to test the McKinnon hypothesis, whether the cost of the credit or quantity matters for private investment.

Investment: The public investment is categorised into infrastructure and non-infrastructure, as suggested in Parker (1995). The sector-wise quarterly data on investment is estimated by maintaining the annual relative shares of private corporate and public investment in each quarter.

Interest rates: The 91-day treasury bills and the 10-year yield on government securities rate for short- and long-term ­interest rates are considered. The ex post real interest rate (backward looking ­retail inflation adjusted) is used for our estimation.

Output gap: Many studies have compared the results of the output gap estimation by using different models in the Indian context. However, despite all the criticisms, this article estimates poten­tial output using the Hodrick-Prescott (HP) filter due to its simplicity.

The simple correlation shown in Figure 2 (p 23) between private and public investment is very high and positive (even with public infrastructure and non-infrastructure), which signals that public inve­stment may not be crowding out the private investment during this period. Non-food credit also indicates a similar picture. The direction of causality, however, may be debated, as some may argue that the decline in non-food credit is ­independent of a lack of demand for ­investment and not vice versa. Interest rates, whether short or long term, highlight a significant positive relationship, suggesting that interest rates matter.

Finally, foreign portfolio capital flows show a low negative relationship with retail inflation dynamics and output gap while a low positive relationship with non-food credit, public and private investment.

Estimation and Results

Our specification Equation (1) incorporates both fiscal policy and monetary policy instruments relevant for encouraging private investment. We consider three models of fiscal instruments (Ipub) as total public investment, public investment in infrastructure, and non-infrastructure separately. We also consider two versions of these three models with monetary policy variable real interest rates (ir) based on short- and long-run interest rates. Since public investment takes time to materialise, our models incorporate regressors for investments made two quarters before the current.

Tables 1 and 2 report our results for models when our iis the short- and long-term interest rate, respectively.
Table 3 provides the confidence intervals under the “meboot” procedure for the first model specifications reported in T­able 1.

Our confidence intervals continue to support “crowding-in” of private investment through public investment for 2011–16. The findings are, thus, consistent with the recent literature ­using Indian data, which does not find crowding-out effects of public investment on private investment.

We find a significantly positive impact of credit cost reductions on corporate ­investment, albeit of a comparatively smaller magnitude than that of increases in public investment.

The direct crowding-in effects of public infrastructure investment on corporate investment evident in the lagged models (see Model 2 results in Tables 1 and 2) signify the spillover or second-round effects of infrastructure investments on economic activity. The instantaneous effect of public infrastructure investment on corporate investment imp­lies a net reduction in project costs of private investment, given the public infrastructure.

A negative coefficient of the output gap indicates a negative impact of macro­economic uncertainties on private corporate investment. The other finding of the study is that the interest rate, that is, cost of credit matters—both short- and long-term. However, the magnitude of the impact is smaller than that of the public investment variable.

During our time period, private investment may have galvanised to attain a larger share of resources, but did not essentially get crowded-out by the mere presence of public sector investment. The significant but opposite signs of non-food credit indicate that a mere quantity of credit may not be enough for enhancing private investment. The direct intervention of the state through focused
infrastructure investment, coupled with the availability of credit, can have a stronger impact on interest rate sensitive private investment. Only by allocating resources for infrastructure, the government encourages private investment.

The negative coefficient of foreign investment (which ideally could be argued to be positive to boost private investment in such equations of investment relations) confirms that the uncertainty towards the stability in the flows of foreign capital had a negative bearing on the scale of private investment.

The initial correlation boxes indicated a high and positive correlation between the two investments. Since we also want to assess the causal directions, we use an exogeneity test statistic (or unanimity index) suggested by Vinod (2017) to determine the direction and strength of causal and exogenous variables.4

Causal paths between variables paired with the private investment are reported in Table 4.

The numbers in the column entitled “corr” of Table 4 are Pearson correlation coefficients. Since all p-values are near zero except for the output gap along line 3 of Table 4, all relations in the table have statistically significantly non-zero Pearson correlation coefficients. However, the symmetry of the matrix of Pearson correlation coefficients means that they cannot suggest anything about the underlying causal directions.

When the value in the “strength” column of Table 4 exceeds 15, the causal direction determination is strong enough to be believed as a preliminary indicator of the true causal direction.

It stands to reason that all variables except LongYld and Ogap along lines 2 and 3 of Table 4 show that long-term yield and output gap influence private investment (PvtInv), but all other variables are sensitive to independent variation in PvtInv data generating process (DGP).

Conclusions

In this article, we used a meboot metho­dology that allows overcoming the unit root and structural change pretests while ruling out the need for any transformations of original time series.

Our causal path analysis using the R package “general Corr” shows that private investment as a DGP has an independent variation that drives the variation in public infra­structure and non-infrastructure investments and the variation in long-term government bond rates. It highlights the importance of private investment as a driving force for the growth of the Indian economy and difficulties in choosing policies to influence it.

Our meboot results indicate evidence in support of “crowding-in” of private ­investment through public investment. We find that public infrastructure inve­stment is significant in determining private investment, and that a low-interest rate encourages private corporate investment.

Private corporate investment is often cyclical, whereby investment booms are followed by recessions, reflecting among other issues the fact that firm-level capa­city utilisation or capacity addition are often bulky, expensive, and uncertain. Our time period covers mostly a recessionary phase of the investment cycle following a modest expansion. In the ­absence of data to cover many business cycles, we capture some aspects of cyclical behaviour by including the “output gap” variable in the model. The public policy implication of our chapter is that the government should remove the infrastructure and bureaucratic bottlenecks in the economy by enhancing “ease of doing business” in India.

Notes

http://www.mospi.gov.in/sites/default/files/press_release/nad_press_release_30jan15.pdf.

2 For further detailed on the seven-step algorithm, see Vinod and Lopez-de-Lacalle (2009); Vinod (2013).

3 See Chakraborty (2007) for detailed derivation of the equation.

4 See Vinod et al (2020) for more details on the application.

References

Bahal, G, M Raissi and V Tulini (2015): “Crowding-out or Crowding-in? Public and Private Investment in India,” IMF Working Paper 264.

Chakraborty, Lekha (2007): “Fiscal Deficit, Capital Formation, and Crowding Out in India: Evidence from an Asymmetric VAR Model,” Economics Working Paper Archive, The Levy Economics Institute, New York, WP 518.

— (2016): Fiscal Consolidation, Budget Deficit and Macroeconomy, New Delhi: Sage Publications. 

Chhibber, A and A Kalloor (2016): “Reviving Private Investment in India: Determinants and Policy Levers,” NIPFP Working paper series, National Institute of Public Finance and Policy, New Delhi, WP 181.

Dash, P (2016): “The Impact of Public Investment on Private Investment: Evidence from India,” Journal for Decision Makers, Vol 41, No 4, pp 288–307.

Efron, B (1979): “Bootstrap Methods: Another Look at Jackknife,” Annals of Statistics, Vol 7, pp 1–26.

Mallick, J (2016): “Effects of Government Investment Shocks on Private Investment and Income in India,” Indian Council for Research on International Economic Relations, New ­Delhi, WP 315.

Parker, K (1995): “The Behaviour of Private Investment,” IMF Occasional Paper 134, Washington, DC: International Monetary Fund.

Vinod, H D (2004): “Ranking mutual funds using unconventional utility theory and stochastic dominance,” Journal of Empirical Finance, Elsevier, Vol 11, No 3, June, pp 353–77.

— (2013): Maximum Entropy Bootstrap Algorithm Enhancements, SSRN eLibrary, http://ssrn.com/paper=2285041.

— (2017): Causal Paths and Exogeneity Tests in GeneralCorr Package for Air Pollution and Monetary Policy, SSRN eLibrary, http://ssrn.com/paper=2982128.

Vinod, H D and Javier Lopez-de-Lacalle (2009): “Maximum Entropy Bootstrap for Time Series: The Meboot R Package,” Journal of Statistical Software, Vol 29, No 5, pp 1–19, http://www.jstatsoft.org/v29/i05/.

Vinod, H D, H Karun and L Chakraborty (2020): “Encouraging Private Corporate Investment in India,” Handbook of Statistics, V Hrishikesh and C R Rao (eds), chapter 5, Financial, Macro and Micro Econometrics Using R, Vol 42, North Holland: Elsevier, pp 155–83.

 

Updated On : 29th Jul, 2020

Comments

(-) Hide

EPW looks forward to your comments. Please note that comments are moderated as per our comments policy. They may take some time to appear. A comment, if suitable, may be selected for publication in the Letters pages of EPW.

Back to Top