- Email: [email protected]
Transmission mechanism of monetary policy in India
Journal of Asian Economics 21 (2010) 186–197
Contents lists available at ScienceDirect
Journal of Asian Economics
Transmission mechanism of monetary policy in India Abdul Aleem * Centre d’e´conomie de l’Universite´ Paris Nord (CEPN), University of Paris XIII, 99 avenue Jean Baptiste Cle´ment 93430 Villetaneuse, France
A R T I C L E I N F O
A B S T R A C T
Article history: Received 6 June 2008 Received in revised form 30 September 2009 Accepted 2 October 2009
This paper examines the transmission mechanism of monetary policy in India. Considering the external constraints on monetary policy, it estimates a series of vector autoregression models to examine the effects of an unanticipated monetary policy tightening on the real sector. The empirical results suggest that the lending rate initially increases in response to a monetary tightening. Banks play an important role in the transmission of monetary policy shocks to the real sector. ß 2009 Elsevier Inc. All rights reserved.
JEL classification: E44 E52 E58 Keywords: Monetary transmission mechanism Bank lending India
1. Introduction Monetary policy affects the real sector at least in the short run, and monetary policy decisions are transmitted to the real sector through different mechanisms. These mechanisms differ from one country to another depending upon their legal and financial structures. Since the beginning of the 1990s, analysis of monetary transmission mechanisms in emerging economies has gained substantial importance due to structural and economic reforms and subsequent transitions to new policy regimes. However, these economies have specific characteristics that differ from those of industrialized countries. Monetary policies in emerging economies are constrained by the world’s major central banks, i.e., the Federal Reserve Bank, the European Central Bank and the Bank of Japan. Hence, the analysis of monetary transmission mechanisms in emerging economies requires a model specification different from that of developed countries. A model misspecification may bias the results. The so-called price-puzzle is one of the consequences of model misspecification (Sims, 1992). Previous empirical studies concerning monetary transmission mechanisms in emerging countries have established the importance of the bank lending channel. However, it is possible that the entire change in aggregate demand after a monetary policy shock occurs via the traditional money channel. Whether the effects of monetary tightening pass through the bank lending channel and not through the traditional money channel remains to be shown. Central banks in emerging economies stabilize exchange rates.1 A flexible exchange rate regime in these economies resembles a de facto peg. Since these economies are characterized by underdeveloped financial markets, their central banks intervene in foreign exchange markets to stabilize exchange rates. This phenomenon is often explained by the hypothesis of ‘‘fear of floating’’ (Calvo & Reinhart, 2000). Given this specific behavior of central banks in emerging economies, a better
* Tel.: +33 149403530; fax: +33 149403334. E-mail address: [email protected]. 1 For more details, see Calvo and Reinhart (2000), Reinhart and Rogoff (2002), Levy-Yeyati and Sturzenegger (2005). 1049-0078/$ – see front matter ß 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.asieco.2009.10.001
A. Aleem / Journal of Asian Economics 21 (2010) 186–197
187
understanding of monetary transmission mechanisms requires an analysis of not only the response of aggregate demand, but also the response of the exchange rate to a monetary policy shock. In the post-reform period, the Reserve Bank of India has adopted market-oriented monetary policy instruments and operating procedures. In the new monetary policy framework, issues related to monetary transmission mechanisms have gained much importance. Some studies have examined specific transmission channels of monetary policy in India.2 Since these studies suffer from the above-mentioned flaws, by taking into account these observations and providing a comprehensive empirical analysis of monetary transmission mechanisms in India, we hope that our work will have important implications for an emerging economy. In this paper, we examine three channels of monetary transmission in India: the bank lending channel, the asset price channel and the exchange rate channel. The paper proceeds as follows. In Section 2, we review the previous work on monetary transmission mechanisms. In Section 3, we propose a benchmark vector autoregression (VAR) model in order to estimate the dynamic responses of GDP, prices and interest rates to an unanticipated monetary policy tightening. In Section 4, we augment the benchmark VAR model to examine the transmission channels of monetary policy and examine the robustness of our results. We conclude in Section 5. 2. Literature review In order to examine the bank lending channel in the United States, Bernanke and Blinder (1988) expanded the standard IS-LM framework by including the bank loans market. Since the beginning of the 1990s, Vector Autoregression (VAR) models have become a widely used tool for analyzing monetary transmission mechanisms. Bernanke and Blinder (1992) examined monetary transmission mechanisms in the United States. They found that monetary policy works partly by affecting the composition of bank assets. Christiano, Eichenbaum, and Evans (1998) showed that the effects of an unanticipated monetary policy shock in the United States are completely transmitted to output, consumption and investment in eighteen months. Peersman and Smets (2001) demonstrated that an unanticipated monetary tightening tends to be followed by a real appreciation of the exchange rate and a temporary fall in output in the euro area. They showed that prices are more sluggish and fall significantly below zero several quarters after the decline in GDP. Morsink and Bayoumi (2001) found that banks play an important role in transmitting monetary shocks to real activity in Japan. Suzuki (2004) discussed the supply versus demand puzzle to examine the credit channel in Japan. He found evidence of the credit channel and showed that an unanticipated monetary policy shock is followed by a permanent increase in land prices. A limited number of empirical studies have examined the monetary transmission mechanisms in emerging economies. Disyatat and Vongsinsirikul (2003) examined the monetary transmission mechanism in Thailand and demonstrated the importance of the bank lending channel. Pandit, Mittal, Roy, and Ghosh (2006) estimated a structural VAR model to examine the bank lending channel in the post-reform period in India. They showed that small banks are more severely affected by monetary tightening than large banks. However, Al-Mashat (2003) found that banks play little role in transmitting monetary policy shocks to the real sector in India. He concluded that the impact of a monetary policy shock on macroeconomic variables is larger after including the exchange rate in the model. An empirical study on monetary transmission in India showed that a positive shock to broad money leads to higher output, while a positive shock to the overnight call money rate produces the opposite effect (Reserve Bank of India, 2003), demonstrating the existence of a narrow credit channel in India. Prasad and Ghosh (2005) examined the relationship between monetary policy and corporate behavior in India. They observed a strengthening of the interest rate channel after 1998. Singh and Kalirajan (2007) concluded that interest rates play an important role in the monetary transmission mechanism in the post-reform period in India. Ahmed, Hastam, Asif, and Yasir (2005) examined different monetary policy channels in Pakistan and demonstrated the importance of the bank lending and interest rate channels. 3. Benchmark model 3.1. Benchmark identification scheme We employ the VAR approach to examine the effects of an unanticipated monetary policy tightening on GDP, prices and overnight call money rate. The VAR approach takes into account the simultaneity between monetary policy variables and the real sector. We identify the benchmark VAR(p) representation as follows: p X
Fi Y ti ¼ QX t þ et
(1)
i¼0
where Yt is the vector of endogenous domestic variables and Xt is the vector of exogenous foreign variables. F and Q are polynomials. et is the vector of structural innovations. The rationale for including the vector of exogenous foreign variables is to take into account external constraints and to control for international economic events. We assume that the exogenous
2
Al-Mashat (2003), Pandit et al. (2006), Reserve Bank of India (2004), Prasad and Ghosh (2005).
A. Aleem / Journal of Asian Economics 21 (2010) 186–197
188
Fig. 1. Spot rates and currency volatility.
variables have contemporaneous effects on the endogenous variables and that there is no feedback effect from endogenous variables to exogenous variables. Monetary policies in emerging economies are constrained by the world’s major central banks, i.e., the Federal Reserve Bank, European Central Bank and the Bank of Japan. Central banks in emerging economies take into account the foreign variables to set the policy rates. These economies are indebted in a foreign currency, e.g., the US dollar or the euro. Default risks can aggravate if the central banks in these economies let the exchange rates fluctuate freely. Similarly, foreign trade in these economies is primarily invoiced either in US dollars or euros. Thus, abrupt exchange rate variations in these economies are harmful for foreign trade. For these reasons, the central banks in emerging economies stabilize exchange rates even though they announce that they do not do so. In order to examine the exchange rate policy of the Reserve Bank of India, we compare the spot rates and volatility of the Indian rupee versus the US dollar and the euro with that of the US dollar versus the euro. The left panel of Fig. 1 depicts the spot rates of the Indian rupee versus the US dollar (INRUSD), the Indian rupee versus the euro (INREUR) and the US dollar versus the euro (USDEUR). We find that INRUSD has remained within the limit of 43–49 rupees per US dollar during the whole period, while INREUR and USDEUR have changed significantly. The right panel of Fig. 1 depicts the currency volatility3 of three spot rates. We find that the INRUSD volatility (VINRUSD) remains very low as compared to the volatilities of other spot rates. If the INRUSD is a fixed rate, every change in the USDEUR will change the INREUR. Hence, INREUR volatility (VINREUR) will be similar to USDEUR volatility (VUSDEUR). In the right panel of Fig. 1, a strong correlation between VINREUR and VUSDEUR suggests the Reserve Bank’s policy to stabilize the Indian rupee vis-a`-vis the US dollar. When a central bank stabilizes the exchange rate, its policy rate follows the foreign interest rate, i.e., the policy rate of the anchor currency’s country. Fig. 2 depicts the evolution of the overnight call money rate and the federal funds rate from 1998 to 2006. The overnight call money rate follows almost the same path as the federal funds rate. Similar movements of the two rates suggest that the Fed’s monetary policy is an external constraint on Indian monetary policy. Therefore, we include the federal funds rate in the vector of exogenous variables. We also include the world commodity prices and the GDP of the United States in the vector of exogenous variables to control for changes in world inflation and demand. Thus, the vector of exogenous foreign variables consists of the world commodity price index (CompiWorld), federal funds rate (ius) and GDP of the United States (yus). h i X 0t ¼ Compiworld ius yus The vector of endogenous domestic variables consists of gross domestic product (GDP), index of domestic prices (Prices) and an indicator of the monetary policy stance (i). Y 0t ¼ ½ GDP
Prices
i
3.2. Data selection In India, the wholesale price index (WPI) is composed of 435 commodities and is available on a weekly basis with a short time lag of two weeks. On the other hand, the consumer price index (CPI) is composed of only 260 commodities and is qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi P 2 We define the currency volatility as the standard deviation, i.e., s ¼ ð nt¼1 ðet eÞ ¯ =n 1Þ, where e is the daily spot rate. We estimate the currency volatility from daily spot rates. In order to calculate the currency volatility, we take the 1st February 1999 as the base year. After calculating the monthly currency volatility, we annualize it. 3
A. Aleem / Journal of Asian Economics 21 (2010) 186–197
189
Fig. 2. Overnight call money rate and federal funds rate.
available on a monthly basis with a time lag of one month. Due to these features of the WPI, it is the most widely used price index in India by both analysts and policy makers to examine developing price trends and is generally considered as an indicator of the inflationary process in the economy. Therefore, we use the WPI as an index of domestic prices. The analysis of monetary transmission mechanisms is sensitive to the choice of interest rate used to capture the monetary policy stance. Since the second half of the 1990s, the monetary transmission mechanism in India has been subject to many changes in the operating procedures of monetary policy with proliferations of new instruments. In April 1997, the bank rate was reactivated as a reference rate to signal the stance of monetary policy. Variations in the bank rate provided signals to commercial banks to modify their deposit and lending rates. In June 2000, the operationalization of the Liquidity Adjustment Facility (LAF) set a corridor for the short-term interest rate through daily repo and reverse repo auctions. The liquidity was absorbed at the reverse repo rate (floor), and the liquidity injection was done at the repo rate (ceiling). Fig. 3 depicts the policy rates and money market rates in India. A significant decline in the volatility of the overnight call money rate since 2000 suggests that the system of LAF enabled the Reserve Bank to modulate short-term liquidity in order to ensure stable conditions in the overnight call money market. In the new monetary framework, the repo rate emerged as the major central bank refinance rate. Beginning March 29, 2004, the entire central bank refinance was made available at the repo rate, culminating in a complete de-linking of standing facilities to banks from bank rate. While the system of LAF was put in place in a phased manner without any disruption in the market, the repo rate emerged as the lending rate of the Reserve Bank, replacing the bank rate for all practical purposes. Since we use the data from 1996Q4 to 2007Q4, the bank rate cannot be used to capture the monetary policy stance. However, if we use the repo rate to capture the monetary policy stance, we are left with a very short time span. Although the policy rates
Fig. 3. Overnight call money rate and the LAF corridor.
A. Aleem / Journal of Asian Economics 21 (2010) 186–197
190
Fig. 4. Structural innovations versus growth in overnight call money rate.
signal the stance of monetary policy, expectations about monetary policy stance are formed through the overnight call money rate. A monetary policy shock affects the short-term interest rates. Following the changes in short-term interest rates, an entire range of monetary and financial variables is affected. Consequently, the changes in the monetary and financial variables affect the aggregate demand. The overnight call money rate is the most closely watched variable in day-to-day conduct of monetary operations. It has served as an informal operating target since the operationalization of the system of LAF. In the context of the extent and timing of intervention in the money market, the overnight call money rate is preferred over other short-term rates to capture the monetary policy stance. Previous empirical studies have used the overnight call money rate to capture the monetary policy stance of the Reserve Bank of India. Singh and Kalirajan (2007) and Al-Mashat (2003) used the overnight call money rate to capture monetary policy stance in order to examine monetary transmission in the post-reform period. Kannan, Sanyal, and Bhoi (2006) used the overnight call money rate as a monetary policy instrument to construct the monetary conditions index for India. Similarly, Virmani (2004) used the overnight call money rate as a monetary policy instrument to estimate the monetary policy rules for India. Hence, we identify an unanticipated shock to the overnight call money rate as an unanticipated monetary policy shock. An unanticipated shock to the overnight call money rate is represented by overnight call money rate innovations, and the responses of other variables to these innovations are represented by the structural impulse responses. In order to examine whether the overnight call money rate innovations correspond to overnight call money rate shocks, we estimate the overnight call money rate innovations in percentage. Fig. 4 depicts the recovered overnight call money rate innovations and percentage quarterly growth of overnight call money rate. We find that the overnight call money rate innovations closely follow the quarterly growth of the overnight call money rate. Hence, the recovered overnight call money rate innovations correspond to the policy actions of the Reserve Bank of India. We perform the ordering of endogenous variables in Eq. (1) by focusing on the dynamic structure of the Indian economy. Monetary policy shocks are orthogonal to the central bank’s information set. Considering the reaction function of the Reserve Bank of India, we argue that the Reserve Bank takes into account the current stage of GDP and prices. Thus, the overnight call money rate responds contemporaneously to shocks to GDP and prices. However, GDP and prices do not respond contemporaneously to overnight call money rate shocks. We use the quarterly seasonally adjusted data from 1996Q4 to 2007Q4. Although some variables appear to be nonstationary, we estimate the VAR model in levels. Sims, Stock, and Watson (1990) demonstrated that a VAR model in levels incurs some loss in estimators’ efficiency but not consistency. The objective of estimating a VAR model in levels is to examine the relationship among variables, not to determine efficient estimates. Previous studies on monetary transmission mechanisms have estimated the VAR models in levels.4 Since GDP and prices are log transformed, their impulse responses to a monetary policy shock are explained as a proportion of the baseline level. The optimal lag length under various criteria seems to be one quarter. However, we feel that this is too short for quarterly data. Hence, we use two lags.5 3.3. Results of the benchmark model Fig. 5 depicts the dynamic responses of GDP, prices and overnight call money rate to a positive one standard deviation overnight call money rate shock. An unanticipated tightening of monetary policy corresponding to a rise of 1.5% in the
4 Favero (2004) also argued that VAR models of monetary transmission mechanisms are very rarely cointegrated VARs. Imposing cointegration restrictions on a VAR model in levels gives more efficient estimators, but at the cost of potential inconsistency due to the imposition of incorrect identifying restrictions. 5 The use of too many lags may create the degrees of freedom problem. Ramasway and Sloek (1997), Morsink and Bayoumi (2001), Al-Mashat (2003), Disyatat and Vongsinsirikul (2003) have used two lags in quarterly estimations of monetary transmission mechanisms in the European Union, Japan, India and Thailand, respectively Ramaswamy and Sloek (1997).
A. Aleem / Journal of Asian Economics 21 (2010) 186–197
191
Fig. 5. Benchmark VAR model: impulse responses to a positive overnight call money rate shock.
overnight call money rate creates a decline in GDP. GDP bottoms out in the third quarter and shows a V-shaped response. Prices also decline in response to a positive overnight call money rate shock. The maximum decline in prices is achieved in the third quarter at 0.1% below the baseline. The price-puzzle disappears after the inclusion of the vector of exogenous variables. The overnight call money rate falls initially to 0.22% below the baseline, but then it converges toward the baseline. This suggests that an unanticipated monetary policy shock has temporary effects on the overnight call money rate. The dynamic impulse responses also show that the prices start declining after the decline in GDP. 4. Monetary transmission channels In order to examine the monetary transmission channels in India, we augment the benchmark VAR model by including a new variable ðwÞ corresponding to the respective monetary transmission channel. We assume that w responds contemporaneously to shocks to GDP, prices and overnight call money rate. However, GDP, prices and overnight call money rate do not respond contemporaneously to a shock to w. Thus, we write the vector of endogenous variables as follows: Y 0t ¼ ½ GDP
prices
i
w
We estimate Eq. (1) to obtain the impulse response of GDP to an unanticipated positive overnight call money rate shock. Then, we exogenize w by treating its lagged values as exogenous variables. Thus, we block off all interactions between w and the other endogenous variables.6 After exogenizing w, the vector of endogenous variables (Y 0t ) is written as follows: Y 0t ¼ ½ GDP
Prices
i
and the vector of exogenous variables is written as follows: h i X 0t ¼ Compiworld ius yus wð1Þ wð pÞ where p represents the number of lags. Thus, by exogenizing w in this way, we obtain a new VAR model similar to the original one. The two VAR models are characterized by similar orthogonalized innovations. The only difference is that in the latter we block off the responses that pass through w. After exogenizing w, we estimate the impulse responses of GDP to a positive overnight call money rate shock. We examine each channel by comparing the impulse responses of GDP in the two VAR models. 4.1. Bank lending channel The bank lending channel relies on two conditions. First, the central bank controls bank lending through its monetary policy instrument. Second, there is no alternative to bank lending, at least for some sectors of borrowers (Barran, Coudert, & Mojon, 1996). Until 1996, the Reserve Bank of India used only the quantity instruments to control the amount of bank loans. Among these instruments were the cash reserve ratio (CRR) and the priority sector lending targets. The objective of priority sector lending targets was to meet the government’s development goals. Since 1997, the Reserve Bank of India has used the price instruments to control indirectly the amount of bank loans. However, the Reserve Bank still uses the priority sector lending targets. India is a bank-based economy characterized by a predominance of bank financing. Since 2005, the bank credit to the commercial sector has accounted for more than 70% of total domestic credit. Fig. 6 shows a comparison of bank credit to
6 Morsink and Bayoumi (2001) and Disyatat and Vongsinsirikul (2003) have used this methodology to examine the monetary transmission mechanisms in Japan and Thailand, respectively. Since we include the vector of exogenous foreign variables in our VAR model, our identification scheme differs from theirs.
192
A. Aleem / Journal of Asian Economics 21 (2010) 186–197
Fig. 6. Indicators of bank lending.
Fig. 7. Currency to deposit ratio.
commercial sector and domestic credit in India. The bank credit to commercial sector as percentage of GDP has increased from 0.6% in 1998 to 1.2% in 2007. This suggests an increasing importance of the banking sector in the Indian economy. The importance of bank lending in the financial system represents the lack of alternative sources of funding for the private sector. Fig. 7 depicts the currency to deposit ratio7 in India. This ratio declined from 10.9 in 1999 to 8.7 in 2007. A gradual decline in the currency to deposit ratio since 1999 suggests an increasing role of banks in financial intermediation. In the bank lending channel, our focus is on the magnification of the effects of an unanticipated monetary policy shock that passes through bank loans. In the standard IS-LM framework, a monetary policy tightening reduces the supply of deposits. A shortage of the supply of deposits creates an increase in the interest rate on bonds. Consequently, an inward movement of the LM curve results in a decline in investment and aggregate demand. This mechanism is commonly known as the money channel. However, in the bank lending channel the effects of a monetary policy tightening are further propagated through changes in bank lending. In the bank lending channel, the interest rates on bonds and loans determine investment and aggregate demand. Thus, a monetary policy tightening reduces not only the supply of deposits, but also the supply of credit. A shortage of the supply of credit creates an increase in the interest rate on loans. Consequently, there is a further decline in investment and aggregate demand. Thus, the response of GDP to a monetary policy tightening is larger in the bank lending channel than in the money channel. In order to examine the importance of the bank lending channel, we proceed in three steps. First, we examine the response of the interest rate on loans to a monetary policy tightening. Second, we examine the effects of a monetary policy tightening on bank loans. Third, given that a monetary policy tightening reduces the aggregate demand, we examine how much of the effects of monetary policy tightening passes through bank lending. The third step tells us about not only the presence of the bank lending channel, but also its importance in the propagation of monetary policy shocks.
7
We define the currency to deposit ratio as the ratio of currency in circulation to demand deposits plus time deposits held by the banks.
A. Aleem / Journal of Asian Economics 21 (2010) 186–197
193
Fig. 8. Response of prime lending rate to a positive overnight call money rate shock.
We examine the effects of a positive overnight call money rate shock on the prime lending rate (PLrate) by using the following vector of endogenous variables: Y 0t ¼ ½ GDP
Prices
i
PLrate
Fig. 8 depicts the dynamic responses of the prime lending rate to an unanticipated positive overnight call money rate shock. The prime lending rate responds immediately to an overnight call money rate shock. A positive overnight call money rate shock creates an initial increase in the prime lending rate to 0.24% above the baseline. After the second quarter, it converges toward the baseline. In order to examine the effects of monetary policy tightening on bank loans and GDP in the bank lending channel, we include the bank credit to the commercial sector. Thus, the vector of endogenous variables consists of GDP, prices, overnight call money rate and bank credit to commercial sector (loans): Y 0t ¼ ½ GDP
Prices
i
Loans
Fig. 9 depicts the dynamic responses of bank loans, prices and GDP to an unanticipated positive overnight call money rate shock. The quantity of bank loans to the commercial sector decreases initially to 0.57% below the baseline in response to a monetary policy tightening. Prices show a similar response to a monetary policy tightening. The solid line in the right panel of Fig. 9 represents the impulse responses of GDP to positive overnight call money rate innovations in the bank lending channel. GDP bottoms out in the third quarter at 0.27% below the baseline. In order to calibrate the importance of the bank lending channel in the transmission of monetary policy shocks, we re-estimate the model after exogenizing the bank loans. After exogenizing the bank loans, the model represents the traditional money channel where there is no role of bank lending and the monetary policy shocks are transmitted to the real sector in the standard IS-LM framework. The dashed line in the right panel of Fig. 9 represents the response of GDP to positive overnight call money rate innovations after exogenizing the bank loans. The response of GDP to positive overnight call money rate innovations is significantly reduced after exogenizing the bank loans. At the beginning of the second year, 70% of the impact of monetary policy tightening comes from bank loans. When we blocked off the channel, the accumulated response of GDP was reduced by 66% in twelve quarters. This difference between the two responses of GDP to positive overnight call money rate innovations suggests the importance of the bank lending channel in India.
Fig. 9. Bank lending channel: impulse responses to a positive overnight call money rate shock.
A. Aleem / Journal of Asian Economics 21 (2010) 186–197
194
Fig. 10. Market capitalization of listed companies (% of GDP).
4.2. Asset price channel Monetary policy shocks are transmitted to asset prices, which subsequently affect the GDP and prices. The degree of capital market development in a country can be examined by observing different parameters such as market capitalization of listed companies, listed stocks and trading volume. Fig. 10 shows that after 2003, the market capitalization of listed companies increased significantly in India. However, it remains lower as compared to developed countries. This suggests that the capital markets in India are not sufficiently developed and that the asset price channel is not significant. We examine the asset price channel by using the following vector of endogenous variables. Y 0t ¼ ½ GDP
Prices
i
SEI
where SEI is the stock exchange index. We use the SENSEX-30 as an index of stock exchange. SENSEX-30 is the index of the Bombay Stock Exchange. It is a widely reported index in both domestic and international markets. It is a basket of thirty constituent stocks representing a sample of large liquid and representative companies. Fig. 11 depicts the impulse response of GDP to a positive overnight call money rate shock. A positive overnight call money rate shock creates a decline in GDP. GDP bottoms out in the fourth quarter at 0.19% below the baseline. When we exogenize the SENSEX-30, GDP shows almost a similar response to a positive overnight call money rate shock. At the beginning of the second year, only 25% of the effects of monetary policy tightening pass through the asset prices. After blocking off the channel, the accumulated response of GDP is reduced by only 24% in twelve quarters. These results suggest that the asset price channel is not important in the transmission of monetary shocks to the real sector in India. 4.3. Exchange rate channel The importance of the exchange rate channel in the transmission of monetary shocks depends on the nature of the exchange rate regime and the degree of openness of the economy. In order to examine the exchange rate channel, we divide the response of the real sector to an interest rate shock into two steps. The first step describes the reaction of the exchange
Fig. 11. Asset price channel: impulse responses to a positive overnight call money rate shock.
A. Aleem / Journal of Asian Economics 21 (2010) 186–197
195
Fig. 12. Exchange rate channel: impulse responses to a positive overnight call money rate shock.
rate to an interest rate shock. The reaction of the exchange rate to an interest rate shock depends on the nature of the exchange rate regime. In a free floating exchange rate regime, the exchange rate reacts more pronouncedly to an interest rate shock. However, if the exchange rate is pegged or heavily managed, it will not respond to the interest rate shock. The second step describes the reaction of the economy to variations in the exchange rate. The response of the real sector to variations in the exchange rate depends positively on the degree of openness of an economy. Since 1993, the official exchange rate regime in India has been the market-determined exchange rate regime. However, frequent interventions by the Reserve Bank of India to stabilize the movements of the Indian rupee vis-a`-vis the US dollar show that the de facto exchange rate regime is different from the de jure exchange rate regime and the Indian rupee seems to be pegged to the US dollar.8 According to the IMF classification of de facto exchange rate regimes in 2006, India has a de facto managed floating regime with an unannounced path of exchange rate (International Monetary Fund, 2006). In 2006, the committee on fuller capital account convertibility affirmed the recommendations of the 1997 committee to the Reserve Bank of India, recommending that the Reserve Bank should maintain a monitoring band of 5% around the real effective exchange rate and should intervene as and when the real effective exchange rate moves outside of this band (Reserve Bank of India, 2006). According to these recommendations, the Reserve Bank of India can use its judgment to intervene even within the band to preclude speculative forces and unwarranted volatility. The committee further recommended that the Reserve Bank of India should undertake a periodic review of the real effective exchange rate, which can be changed as warranted by fundamentals. Moreover, ensuring orderly conditions in the foreign exchange market to avoid excessive exchange rate volatility is one of the objectives of Indian monetary policy. Frequent interventions by the Reserve Bank of India to stabilize the exchange rate weaken the exchange rate channel. Moreover, the large size of the Indian domestic market as compared to its total exports or imports also suggests that the exchange rate channel is not important in the transmission of monetary policy shocks to the real sector. In order to examine the exchange rate channel, we focus on the effects of a positive overnight call money rate shock on GDP that passes through the exchange rate. The vector of endogenous variables consists of GDP, prices, overnight call money rate and real effective exchange rate (REER). Y 0t ¼ ½ GDP
Prices
i
REER
Fig. 12 depicts the responses of GDP, prices and real effective exchange rate to positive innovations in the overnight call money rate and real effective exchange rate. In response to an unanticipated increase in the overnight call money rate, the real effective exchange rate appreciates initially and shows a short-lived reaction to a positive overnight call money rate shock. The effects of a positive overnight call money rate shock on the real effective exchange rate almost disappear in the sixth quarter. A weak response of the real effective exchange rate to a positive monetary policy shock weakens the first step in the exchange rate channel. These results also suggest that the Reserve Bank of India tends to stabilize the real effective exchange rate. The right panel of Fig. 12 shows the responses of GDP to overnight call money rate innovations with and without real effective exchange rate exogenized. In the two cases, the GDP responds to positive overnight call money rate innovations almost in a similar way. These results suggest the absence of an exchange rate channel in India. 4.4. Robustness of results The empirical results in the augmented VAR models suggest the importance of the bank lending channel in India. In order to check the robustness of these empirical results, we examined their statistical significance. We estimated 2S.E. confidence intervals after blocking off each channel. Fig. 13 depicts the impulse responses of GDP to a positive overnight call money rate shock within the channel and after blocking off the channel with 2S.E. confidence intervals.
8
Similarly, Reinhart and Rogoff (2002) found that from July 1995 to December 2001, India had a de facto crawling peg to the US dollar.
196
A. Aleem / Journal of Asian Economics 21 (2010) 186–197
Fig. 13. Response of GDP to a positive overnight call money rate shock.
We find that the responses of GDP to a positive overnight call money rate shock are statistically significant in the bank lending channel. However, GDP shows similar responses to a positive monetary policy shock in the asset price and exchange rate channels. These estimates suggest the robustness of our empirical results and validate the importance of the bank lending channel in India. 5. Conclusion The existence of external constraints on monetary policy in emerging economies requires a model specification different from that of developed countries. This paper provides a comprehensive empirical analysis of the monetary transmission mechanism in India. We estimated a series of VAR models to examine three transmission channels of monetary policy. The benchmark VAR model is composed of a vector of endogenous domestic variables and a vector of exogenous foreign variables. We imposed restrictions on the contemporaneous effects of endogenous variables to have an exact identification of the benchmark VAR model. The results of the benchmark VAR model suggest that an unanticipated monetary policy shock has transitory effects on the overnight call money rate. The price-puzzle vanished after the inclusion of the vector of exogenous foreign variables. Prices and GDP decline after an unanticipated positive overnight call money rate shock. Moreover, prices start declining after a decline in GDP. The Indian economy is a bank-based economy. Since 2005, bank credit to the commercial sector has accounted for more than 70% of total domestic credit. The currency to deposit ratio has declined continuously since 1999. These facts suggest that the banks play an important role in financial intermediation and that the non-financial sector lacks alternative sources of funding. Our empirical results support the importance of the bank lending channel in the transmission of monetary policy shocks to the real sector. The lower market capitalization of listed companies in India as compared to developed countries suggests that the capital markets in India are not sufficiently developed. Empirical estimates in the augmented VAR model suggest that the asset price channel is not important in the transmission of monetary policy shocks to the real sector in India. Central banks in emerging economies stabilize exchange rates even though they announce that they do not do so. Massive interventions by the Reserve Bank of India in the foreign exchange market to stabilize the exchange rate weaken the exchange rate channel. The shortlived response of the real effective exchange rate to an unanticipated monetary policy tightening suggests that the exchange rate channel is not important in the transmission of monetary policy shocks in India. This analysis provides some important theoretical and policy implications. First, Indian monetary policy is constrained by the Fed’s monetary policy. Hence, an analysis of Indian monetary policy requires the inclusion of the federal funds rate in the information set of the Reserve Bank of India. A proper model specification, considering the external constraints on monetary policy and controlling for international economic events, reduces the bias. Second, the Reserve Bank of India intervenes massively in the foreign exchange market to stabilize the exchange rate. The Indian rupee seems to be pegged to the US dollar. Hence, a proper comprehension of the monetary transmission mechanism in India requires the analysis not only of the response of GDP, but also of the response of the exchange rate to a monetary policy shock. Third, banks play an important role in financial intermediation in the Indian economy, and their strong representation reflects the lack of alternative sources of funding for the private sector. Appendix A. Data sources Overnight call money rate, bank rate, repo rate, reverse repo rate, bank credit to commercial sector, currency in circulation, time deposits, demand deposits: Reserve Bank of India; and Reuters. Wholesale Price Index, federal funds rate, GDP of the United States and India: International Monetary Fund, International Financial Statistics.
A. Aleem / Journal of Asian Economics 21 (2010) 186–197
197
Sensitive index (BSE-30), spot rates: Reuters; and Eurostat. Real effective exchange rate: The Bank for International Settlements. Prime lending rate: OECD. World commodity price index: The economist. Market capitalization of listed companies: World Bank, World Development Indicators. References Ahmed, N., Hastam, S., Asif, A. I., & Yasir, M. A. (2005). Transmission mechanism of monetary policy in Pakistan. SBP research bulletin, Vol. 1, No. 1, State Bank of Pakistan. Al-Mashat, R. (2003). Monetary policy transmission in India: Selected issues and statistical appendix. Country report no. 03/261. International Monetary Fund. Barran, F., Coudert, V., & Mojon, B. (1996). The transmission of monetary policy in European countries. Working paper no. 96-03. Paris: CEPII. Bernanke, B., & Blinder, A. (1988). Credit, money and aggregated demand. The American economic review, papers and proceedings of the one-hundredth annual meeting of the American economic association, Vol. 78 (pp. 435–439). Bernanke, B., & Blinder, A. (1992). The Federal funds rate and the channels of monetary transmission. The American Economic Review, 82, 901–921. Calvo, G., & Reinhart, C. (2000). Fear of floating. NBER working paper no. 7993. Cambridge: National Bureau of Economic Research. Christiano, L. J., Eichenbaum, M., & Evans, C. (1998). Monetary policy shocks: What have we learned and to what end? NBER working paper no. 6400. Cambridge: National Bureau of Economic Research. Disyatat, P., & Vongsinsirikul, P. (2003). Monetary policy and the transmission mechanism in Thailand. Journal of Asian economics, 4, 389–418. Favero, C. A. (2004). Applied macroeconometrics. Oxford University Press. pp. 162–211. International Monetary Fund. (2006, July). De facto classification of exchange rate regimes and monetary policy framework. IMF. Available on http://www.imf.org/ external/np/mfd/er/2006/eng/0706.htm. Kannan, R., Sanyal, S., & Bhoi, B. (2006, Winter). Monetary conditions index for India. Reserve Bank of India occasional papers 27, no. 3. Levy-Yeyati, E., & Sturzenegger, F. (2005). Classifying exchange rate regimes: Deeds vs. words. European Economic Review, 49, 1603–1635. Morsink, J., & Bayoumi, T. (2001). A peek inside the black box: The monetary transmission mechanism in Japan. IMF staff papers, Vol. 48, International Monetary Fund. pp. 22–57. Pandit, B. L., Mittal, A., Roy, M., & Ghosh, S. (2006). Transmission of monetary policy and the bank lending channel: Analysis and evidence for India. Study number 25, Mumbai: Development Research Group, Department of Economic Analysis And Policy, Reserve Bank of India. Peersman, G., & Smets, F. (2001). The monetary transmission mechanism in the euro area: More evidence from VAR analysis. Working paper no. 91. Germany: European Central Bank. Prasad, A., & Ghosh, S. (2005). Monetary policy and corporate behaviour in India. IMF working paper no. 05/25. International Monetary Fund. Ramaswamy, R., & Sloek, T. (1997). The real effects of monetary policy in the European Union: What are the differences? IMF working paper 97/160. Reinhart, C., & Rogoff, K. (2002). Modern history of exchange rate arrangements: A reinterpretation. NBER working paper no. 8963. Cambridge: National Bureau of Economic Research. Reserve Bank of India. (2003, March). Money, credit and prices. Report on currency and finance. Mumbai: Reserve Bank of India. Reserve Bank of India. (2004, December). Monetary transmission mechanism. Report on currency and finance 2003-04 (pp. 171–193). Mumbai: Reserve Bank of India. Reserve Bank of India. (2006, July). Chapter 5: Interaction of monetary policy and exchange rate policy. Report of the committee on fuller capital account convertibility. Sims, C. (1992). Interpreting the macro-economic time series facts: The effects of monetary policy. European Economic Review, 36, 975–1011. Sims, C., Stock, J. H., & Watson, M. W. (1990). Inference in linear time series models with some unit roots. Econometrica, 58, 113–144. Singh, K., & Kalirajan, K. (2007). Monetary transmission in post-reform India: An evaluation. Journal of Asia pacific economy, 12, 158–187. Suzuki, T. (2004). Credit channel of monetary policy in Japan: Resolving the supply versus demand puzzle. Applied Economics, 36, 2385–2396. Virmani, V. (2004). Operationalizing Taylor-type rules for the Indian economy: Issues and some results (1992Q–2001Q1). IIMA working paper no. 2004-07-04.
Contents lists available at ScienceDirect
Journal of Asian Economics
Transmission mechanism of monetary policy in India Abdul Aleem * Centre d’e´conomie de l’Universite´ Paris Nord (CEPN), University of Paris XIII, 99 avenue Jean Baptiste Cle´ment 93430 Villetaneuse, France
A R T I C L E I N F O
A B S T R A C T
Article history: Received 6 June 2008 Received in revised form 30 September 2009 Accepted 2 October 2009
This paper examines the transmission mechanism of monetary policy in India. Considering the external constraints on monetary policy, it estimates a series of vector autoregression models to examine the effects of an unanticipated monetary policy tightening on the real sector. The empirical results suggest that the lending rate initially increases in response to a monetary tightening. Banks play an important role in the transmission of monetary policy shocks to the real sector. ß 2009 Elsevier Inc. All rights reserved.
JEL classification: E44 E52 E58 Keywords: Monetary transmission mechanism Bank lending India
1. Introduction Monetary policy affects the real sector at least in the short run, and monetary policy decisions are transmitted to the real sector through different mechanisms. These mechanisms differ from one country to another depending upon their legal and financial structures. Since the beginning of the 1990s, analysis of monetary transmission mechanisms in emerging economies has gained substantial importance due to structural and economic reforms and subsequent transitions to new policy regimes. However, these economies have specific characteristics that differ from those of industrialized countries. Monetary policies in emerging economies are constrained by the world’s major central banks, i.e., the Federal Reserve Bank, the European Central Bank and the Bank of Japan. Hence, the analysis of monetary transmission mechanisms in emerging economies requires a model specification different from that of developed countries. A model misspecification may bias the results. The so-called price-puzzle is one of the consequences of model misspecification (Sims, 1992). Previous empirical studies concerning monetary transmission mechanisms in emerging countries have established the importance of the bank lending channel. However, it is possible that the entire change in aggregate demand after a monetary policy shock occurs via the traditional money channel. Whether the effects of monetary tightening pass through the bank lending channel and not through the traditional money channel remains to be shown. Central banks in emerging economies stabilize exchange rates.1 A flexible exchange rate regime in these economies resembles a de facto peg. Since these economies are characterized by underdeveloped financial markets, their central banks intervene in foreign exchange markets to stabilize exchange rates. This phenomenon is often explained by the hypothesis of ‘‘fear of floating’’ (Calvo & Reinhart, 2000). Given this specific behavior of central banks in emerging economies, a better
* Tel.: +33 149403530; fax: +33 149403334. E-mail address: [email protected]. 1 For more details, see Calvo and Reinhart (2000), Reinhart and Rogoff (2002), Levy-Yeyati and Sturzenegger (2005). 1049-0078/$ – see front matter ß 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.asieco.2009.10.001
A. Aleem / Journal of Asian Economics 21 (2010) 186–197
187
understanding of monetary transmission mechanisms requires an analysis of not only the response of aggregate demand, but also the response of the exchange rate to a monetary policy shock. In the post-reform period, the Reserve Bank of India has adopted market-oriented monetary policy instruments and operating procedures. In the new monetary policy framework, issues related to monetary transmission mechanisms have gained much importance. Some studies have examined specific transmission channels of monetary policy in India.2 Since these studies suffer from the above-mentioned flaws, by taking into account these observations and providing a comprehensive empirical analysis of monetary transmission mechanisms in India, we hope that our work will have important implications for an emerging economy. In this paper, we examine three channels of monetary transmission in India: the bank lending channel, the asset price channel and the exchange rate channel. The paper proceeds as follows. In Section 2, we review the previous work on monetary transmission mechanisms. In Section 3, we propose a benchmark vector autoregression (VAR) model in order to estimate the dynamic responses of GDP, prices and interest rates to an unanticipated monetary policy tightening. In Section 4, we augment the benchmark VAR model to examine the transmission channels of monetary policy and examine the robustness of our results. We conclude in Section 5. 2. Literature review In order to examine the bank lending channel in the United States, Bernanke and Blinder (1988) expanded the standard IS-LM framework by including the bank loans market. Since the beginning of the 1990s, Vector Autoregression (VAR) models have become a widely used tool for analyzing monetary transmission mechanisms. Bernanke and Blinder (1992) examined monetary transmission mechanisms in the United States. They found that monetary policy works partly by affecting the composition of bank assets. Christiano, Eichenbaum, and Evans (1998) showed that the effects of an unanticipated monetary policy shock in the United States are completely transmitted to output, consumption and investment in eighteen months. Peersman and Smets (2001) demonstrated that an unanticipated monetary tightening tends to be followed by a real appreciation of the exchange rate and a temporary fall in output in the euro area. They showed that prices are more sluggish and fall significantly below zero several quarters after the decline in GDP. Morsink and Bayoumi (2001) found that banks play an important role in transmitting monetary shocks to real activity in Japan. Suzuki (2004) discussed the supply versus demand puzzle to examine the credit channel in Japan. He found evidence of the credit channel and showed that an unanticipated monetary policy shock is followed by a permanent increase in land prices. A limited number of empirical studies have examined the monetary transmission mechanisms in emerging economies. Disyatat and Vongsinsirikul (2003) examined the monetary transmission mechanism in Thailand and demonstrated the importance of the bank lending channel. Pandit, Mittal, Roy, and Ghosh (2006) estimated a structural VAR model to examine the bank lending channel in the post-reform period in India. They showed that small banks are more severely affected by monetary tightening than large banks. However, Al-Mashat (2003) found that banks play little role in transmitting monetary policy shocks to the real sector in India. He concluded that the impact of a monetary policy shock on macroeconomic variables is larger after including the exchange rate in the model. An empirical study on monetary transmission in India showed that a positive shock to broad money leads to higher output, while a positive shock to the overnight call money rate produces the opposite effect (Reserve Bank of India, 2003), demonstrating the existence of a narrow credit channel in India. Prasad and Ghosh (2005) examined the relationship between monetary policy and corporate behavior in India. They observed a strengthening of the interest rate channel after 1998. Singh and Kalirajan (2007) concluded that interest rates play an important role in the monetary transmission mechanism in the post-reform period in India. Ahmed, Hastam, Asif, and Yasir (2005) examined different monetary policy channels in Pakistan and demonstrated the importance of the bank lending and interest rate channels. 3. Benchmark model 3.1. Benchmark identification scheme We employ the VAR approach to examine the effects of an unanticipated monetary policy tightening on GDP, prices and overnight call money rate. The VAR approach takes into account the simultaneity between monetary policy variables and the real sector. We identify the benchmark VAR(p) representation as follows: p X
Fi Y ti ¼ QX t þ et
(1)
i¼0
where Yt is the vector of endogenous domestic variables and Xt is the vector of exogenous foreign variables. F and Q are polynomials. et is the vector of structural innovations. The rationale for including the vector of exogenous foreign variables is to take into account external constraints and to control for international economic events. We assume that the exogenous
2
Al-Mashat (2003), Pandit et al. (2006), Reserve Bank of India (2004), Prasad and Ghosh (2005).
A. Aleem / Journal of Asian Economics 21 (2010) 186–197
188
Fig. 1. Spot rates and currency volatility.
variables have contemporaneous effects on the endogenous variables and that there is no feedback effect from endogenous variables to exogenous variables. Monetary policies in emerging economies are constrained by the world’s major central banks, i.e., the Federal Reserve Bank, European Central Bank and the Bank of Japan. Central banks in emerging economies take into account the foreign variables to set the policy rates. These economies are indebted in a foreign currency, e.g., the US dollar or the euro. Default risks can aggravate if the central banks in these economies let the exchange rates fluctuate freely. Similarly, foreign trade in these economies is primarily invoiced either in US dollars or euros. Thus, abrupt exchange rate variations in these economies are harmful for foreign trade. For these reasons, the central banks in emerging economies stabilize exchange rates even though they announce that they do not do so. In order to examine the exchange rate policy of the Reserve Bank of India, we compare the spot rates and volatility of the Indian rupee versus the US dollar and the euro with that of the US dollar versus the euro. The left panel of Fig. 1 depicts the spot rates of the Indian rupee versus the US dollar (INRUSD), the Indian rupee versus the euro (INREUR) and the US dollar versus the euro (USDEUR). We find that INRUSD has remained within the limit of 43–49 rupees per US dollar during the whole period, while INREUR and USDEUR have changed significantly. The right panel of Fig. 1 depicts the currency volatility3 of three spot rates. We find that the INRUSD volatility (VINRUSD) remains very low as compared to the volatilities of other spot rates. If the INRUSD is a fixed rate, every change in the USDEUR will change the INREUR. Hence, INREUR volatility (VINREUR) will be similar to USDEUR volatility (VUSDEUR). In the right panel of Fig. 1, a strong correlation between VINREUR and VUSDEUR suggests the Reserve Bank’s policy to stabilize the Indian rupee vis-a`-vis the US dollar. When a central bank stabilizes the exchange rate, its policy rate follows the foreign interest rate, i.e., the policy rate of the anchor currency’s country. Fig. 2 depicts the evolution of the overnight call money rate and the federal funds rate from 1998 to 2006. The overnight call money rate follows almost the same path as the federal funds rate. Similar movements of the two rates suggest that the Fed’s monetary policy is an external constraint on Indian monetary policy. Therefore, we include the federal funds rate in the vector of exogenous variables. We also include the world commodity prices and the GDP of the United States in the vector of exogenous variables to control for changes in world inflation and demand. Thus, the vector of exogenous foreign variables consists of the world commodity price index (CompiWorld), federal funds rate (ius) and GDP of the United States (yus). h i X 0t ¼ Compiworld ius yus The vector of endogenous domestic variables consists of gross domestic product (GDP), index of domestic prices (Prices) and an indicator of the monetary policy stance (i). Y 0t ¼ ½ GDP
Prices
i
3.2. Data selection In India, the wholesale price index (WPI) is composed of 435 commodities and is available on a weekly basis with a short time lag of two weeks. On the other hand, the consumer price index (CPI) is composed of only 260 commodities and is qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi P 2 We define the currency volatility as the standard deviation, i.e., s ¼ ð nt¼1 ðet eÞ ¯ =n 1Þ, where e is the daily spot rate. We estimate the currency volatility from daily spot rates. In order to calculate the currency volatility, we take the 1st February 1999 as the base year. After calculating the monthly currency volatility, we annualize it. 3
A. Aleem / Journal of Asian Economics 21 (2010) 186–197
189
Fig. 2. Overnight call money rate and federal funds rate.
available on a monthly basis with a time lag of one month. Due to these features of the WPI, it is the most widely used price index in India by both analysts and policy makers to examine developing price trends and is generally considered as an indicator of the inflationary process in the economy. Therefore, we use the WPI as an index of domestic prices. The analysis of monetary transmission mechanisms is sensitive to the choice of interest rate used to capture the monetary policy stance. Since the second half of the 1990s, the monetary transmission mechanism in India has been subject to many changes in the operating procedures of monetary policy with proliferations of new instruments. In April 1997, the bank rate was reactivated as a reference rate to signal the stance of monetary policy. Variations in the bank rate provided signals to commercial banks to modify their deposit and lending rates. In June 2000, the operationalization of the Liquidity Adjustment Facility (LAF) set a corridor for the short-term interest rate through daily repo and reverse repo auctions. The liquidity was absorbed at the reverse repo rate (floor), and the liquidity injection was done at the repo rate (ceiling). Fig. 3 depicts the policy rates and money market rates in India. A significant decline in the volatility of the overnight call money rate since 2000 suggests that the system of LAF enabled the Reserve Bank to modulate short-term liquidity in order to ensure stable conditions in the overnight call money market. In the new monetary framework, the repo rate emerged as the major central bank refinance rate. Beginning March 29, 2004, the entire central bank refinance was made available at the repo rate, culminating in a complete de-linking of standing facilities to banks from bank rate. While the system of LAF was put in place in a phased manner without any disruption in the market, the repo rate emerged as the lending rate of the Reserve Bank, replacing the bank rate for all practical purposes. Since we use the data from 1996Q4 to 2007Q4, the bank rate cannot be used to capture the monetary policy stance. However, if we use the repo rate to capture the monetary policy stance, we are left with a very short time span. Although the policy rates
Fig. 3. Overnight call money rate and the LAF corridor.
A. Aleem / Journal of Asian Economics 21 (2010) 186–197
190
Fig. 4. Structural innovations versus growth in overnight call money rate.
signal the stance of monetary policy, expectations about monetary policy stance are formed through the overnight call money rate. A monetary policy shock affects the short-term interest rates. Following the changes in short-term interest rates, an entire range of monetary and financial variables is affected. Consequently, the changes in the monetary and financial variables affect the aggregate demand. The overnight call money rate is the most closely watched variable in day-to-day conduct of monetary operations. It has served as an informal operating target since the operationalization of the system of LAF. In the context of the extent and timing of intervention in the money market, the overnight call money rate is preferred over other short-term rates to capture the monetary policy stance. Previous empirical studies have used the overnight call money rate to capture the monetary policy stance of the Reserve Bank of India. Singh and Kalirajan (2007) and Al-Mashat (2003) used the overnight call money rate to capture monetary policy stance in order to examine monetary transmission in the post-reform period. Kannan, Sanyal, and Bhoi (2006) used the overnight call money rate as a monetary policy instrument to construct the monetary conditions index for India. Similarly, Virmani (2004) used the overnight call money rate as a monetary policy instrument to estimate the monetary policy rules for India. Hence, we identify an unanticipated shock to the overnight call money rate as an unanticipated monetary policy shock. An unanticipated shock to the overnight call money rate is represented by overnight call money rate innovations, and the responses of other variables to these innovations are represented by the structural impulse responses. In order to examine whether the overnight call money rate innovations correspond to overnight call money rate shocks, we estimate the overnight call money rate innovations in percentage. Fig. 4 depicts the recovered overnight call money rate innovations and percentage quarterly growth of overnight call money rate. We find that the overnight call money rate innovations closely follow the quarterly growth of the overnight call money rate. Hence, the recovered overnight call money rate innovations correspond to the policy actions of the Reserve Bank of India. We perform the ordering of endogenous variables in Eq. (1) by focusing on the dynamic structure of the Indian economy. Monetary policy shocks are orthogonal to the central bank’s information set. Considering the reaction function of the Reserve Bank of India, we argue that the Reserve Bank takes into account the current stage of GDP and prices. Thus, the overnight call money rate responds contemporaneously to shocks to GDP and prices. However, GDP and prices do not respond contemporaneously to overnight call money rate shocks. We use the quarterly seasonally adjusted data from 1996Q4 to 2007Q4. Although some variables appear to be nonstationary, we estimate the VAR model in levels. Sims, Stock, and Watson (1990) demonstrated that a VAR model in levels incurs some loss in estimators’ efficiency but not consistency. The objective of estimating a VAR model in levels is to examine the relationship among variables, not to determine efficient estimates. Previous studies on monetary transmission mechanisms have estimated the VAR models in levels.4 Since GDP and prices are log transformed, their impulse responses to a monetary policy shock are explained as a proportion of the baseline level. The optimal lag length under various criteria seems to be one quarter. However, we feel that this is too short for quarterly data. Hence, we use two lags.5 3.3. Results of the benchmark model Fig. 5 depicts the dynamic responses of GDP, prices and overnight call money rate to a positive one standard deviation overnight call money rate shock. An unanticipated tightening of monetary policy corresponding to a rise of 1.5% in the
4 Favero (2004) also argued that VAR models of monetary transmission mechanisms are very rarely cointegrated VARs. Imposing cointegration restrictions on a VAR model in levels gives more efficient estimators, but at the cost of potential inconsistency due to the imposition of incorrect identifying restrictions. 5 The use of too many lags may create the degrees of freedom problem. Ramasway and Sloek (1997), Morsink and Bayoumi (2001), Al-Mashat (2003), Disyatat and Vongsinsirikul (2003) have used two lags in quarterly estimations of monetary transmission mechanisms in the European Union, Japan, India and Thailand, respectively Ramaswamy and Sloek (1997).
A. Aleem / Journal of Asian Economics 21 (2010) 186–197
191
Fig. 5. Benchmark VAR model: impulse responses to a positive overnight call money rate shock.
overnight call money rate creates a decline in GDP. GDP bottoms out in the third quarter and shows a V-shaped response. Prices also decline in response to a positive overnight call money rate shock. The maximum decline in prices is achieved in the third quarter at 0.1% below the baseline. The price-puzzle disappears after the inclusion of the vector of exogenous variables. The overnight call money rate falls initially to 0.22% below the baseline, but then it converges toward the baseline. This suggests that an unanticipated monetary policy shock has temporary effects on the overnight call money rate. The dynamic impulse responses also show that the prices start declining after the decline in GDP. 4. Monetary transmission channels In order to examine the monetary transmission channels in India, we augment the benchmark VAR model by including a new variable ðwÞ corresponding to the respective monetary transmission channel. We assume that w responds contemporaneously to shocks to GDP, prices and overnight call money rate. However, GDP, prices and overnight call money rate do not respond contemporaneously to a shock to w. Thus, we write the vector of endogenous variables as follows: Y 0t ¼ ½ GDP
prices
i
w
We estimate Eq. (1) to obtain the impulse response of GDP to an unanticipated positive overnight call money rate shock. Then, we exogenize w by treating its lagged values as exogenous variables. Thus, we block off all interactions between w and the other endogenous variables.6 After exogenizing w, the vector of endogenous variables (Y 0t ) is written as follows: Y 0t ¼ ½ GDP
Prices
i
and the vector of exogenous variables is written as follows: h i X 0t ¼ Compiworld ius yus wð1Þ wð pÞ where p represents the number of lags. Thus, by exogenizing w in this way, we obtain a new VAR model similar to the original one. The two VAR models are characterized by similar orthogonalized innovations. The only difference is that in the latter we block off the responses that pass through w. After exogenizing w, we estimate the impulse responses of GDP to a positive overnight call money rate shock. We examine each channel by comparing the impulse responses of GDP in the two VAR models. 4.1. Bank lending channel The bank lending channel relies on two conditions. First, the central bank controls bank lending through its monetary policy instrument. Second, there is no alternative to bank lending, at least for some sectors of borrowers (Barran, Coudert, & Mojon, 1996). Until 1996, the Reserve Bank of India used only the quantity instruments to control the amount of bank loans. Among these instruments were the cash reserve ratio (CRR) and the priority sector lending targets. The objective of priority sector lending targets was to meet the government’s development goals. Since 1997, the Reserve Bank of India has used the price instruments to control indirectly the amount of bank loans. However, the Reserve Bank still uses the priority sector lending targets. India is a bank-based economy characterized by a predominance of bank financing. Since 2005, the bank credit to the commercial sector has accounted for more than 70% of total domestic credit. Fig. 6 shows a comparison of bank credit to
6 Morsink and Bayoumi (2001) and Disyatat and Vongsinsirikul (2003) have used this methodology to examine the monetary transmission mechanisms in Japan and Thailand, respectively. Since we include the vector of exogenous foreign variables in our VAR model, our identification scheme differs from theirs.
192
A. Aleem / Journal of Asian Economics 21 (2010) 186–197
Fig. 6. Indicators of bank lending.
Fig. 7. Currency to deposit ratio.
commercial sector and domestic credit in India. The bank credit to commercial sector as percentage of GDP has increased from 0.6% in 1998 to 1.2% in 2007. This suggests an increasing importance of the banking sector in the Indian economy. The importance of bank lending in the financial system represents the lack of alternative sources of funding for the private sector. Fig. 7 depicts the currency to deposit ratio7 in India. This ratio declined from 10.9 in 1999 to 8.7 in 2007. A gradual decline in the currency to deposit ratio since 1999 suggests an increasing role of banks in financial intermediation. In the bank lending channel, our focus is on the magnification of the effects of an unanticipated monetary policy shock that passes through bank loans. In the standard IS-LM framework, a monetary policy tightening reduces the supply of deposits. A shortage of the supply of deposits creates an increase in the interest rate on bonds. Consequently, an inward movement of the LM curve results in a decline in investment and aggregate demand. This mechanism is commonly known as the money channel. However, in the bank lending channel the effects of a monetary policy tightening are further propagated through changes in bank lending. In the bank lending channel, the interest rates on bonds and loans determine investment and aggregate demand. Thus, a monetary policy tightening reduces not only the supply of deposits, but also the supply of credit. A shortage of the supply of credit creates an increase in the interest rate on loans. Consequently, there is a further decline in investment and aggregate demand. Thus, the response of GDP to a monetary policy tightening is larger in the bank lending channel than in the money channel. In order to examine the importance of the bank lending channel, we proceed in three steps. First, we examine the response of the interest rate on loans to a monetary policy tightening. Second, we examine the effects of a monetary policy tightening on bank loans. Third, given that a monetary policy tightening reduces the aggregate demand, we examine how much of the effects of monetary policy tightening passes through bank lending. The third step tells us about not only the presence of the bank lending channel, but also its importance in the propagation of monetary policy shocks.
7
We define the currency to deposit ratio as the ratio of currency in circulation to demand deposits plus time deposits held by the banks.
A. Aleem / Journal of Asian Economics 21 (2010) 186–197
193
Fig. 8. Response of prime lending rate to a positive overnight call money rate shock.
We examine the effects of a positive overnight call money rate shock on the prime lending rate (PLrate) by using the following vector of endogenous variables: Y 0t ¼ ½ GDP
Prices
i
PLrate
Fig. 8 depicts the dynamic responses of the prime lending rate to an unanticipated positive overnight call money rate shock. The prime lending rate responds immediately to an overnight call money rate shock. A positive overnight call money rate shock creates an initial increase in the prime lending rate to 0.24% above the baseline. After the second quarter, it converges toward the baseline. In order to examine the effects of monetary policy tightening on bank loans and GDP in the bank lending channel, we include the bank credit to the commercial sector. Thus, the vector of endogenous variables consists of GDP, prices, overnight call money rate and bank credit to commercial sector (loans): Y 0t ¼ ½ GDP
Prices
i
Loans
Fig. 9 depicts the dynamic responses of bank loans, prices and GDP to an unanticipated positive overnight call money rate shock. The quantity of bank loans to the commercial sector decreases initially to 0.57% below the baseline in response to a monetary policy tightening. Prices show a similar response to a monetary policy tightening. The solid line in the right panel of Fig. 9 represents the impulse responses of GDP to positive overnight call money rate innovations in the bank lending channel. GDP bottoms out in the third quarter at 0.27% below the baseline. In order to calibrate the importance of the bank lending channel in the transmission of monetary policy shocks, we re-estimate the model after exogenizing the bank loans. After exogenizing the bank loans, the model represents the traditional money channel where there is no role of bank lending and the monetary policy shocks are transmitted to the real sector in the standard IS-LM framework. The dashed line in the right panel of Fig. 9 represents the response of GDP to positive overnight call money rate innovations after exogenizing the bank loans. The response of GDP to positive overnight call money rate innovations is significantly reduced after exogenizing the bank loans. At the beginning of the second year, 70% of the impact of monetary policy tightening comes from bank loans. When we blocked off the channel, the accumulated response of GDP was reduced by 66% in twelve quarters. This difference between the two responses of GDP to positive overnight call money rate innovations suggests the importance of the bank lending channel in India.
Fig. 9. Bank lending channel: impulse responses to a positive overnight call money rate shock.
A. Aleem / Journal of Asian Economics 21 (2010) 186–197
194
Fig. 10. Market capitalization of listed companies (% of GDP).
4.2. Asset price channel Monetary policy shocks are transmitted to asset prices, which subsequently affect the GDP and prices. The degree of capital market development in a country can be examined by observing different parameters such as market capitalization of listed companies, listed stocks and trading volume. Fig. 10 shows that after 2003, the market capitalization of listed companies increased significantly in India. However, it remains lower as compared to developed countries. This suggests that the capital markets in India are not sufficiently developed and that the asset price channel is not significant. We examine the asset price channel by using the following vector of endogenous variables. Y 0t ¼ ½ GDP
Prices
i
SEI
where SEI is the stock exchange index. We use the SENSEX-30 as an index of stock exchange. SENSEX-30 is the index of the Bombay Stock Exchange. It is a widely reported index in both domestic and international markets. It is a basket of thirty constituent stocks representing a sample of large liquid and representative companies. Fig. 11 depicts the impulse response of GDP to a positive overnight call money rate shock. A positive overnight call money rate shock creates a decline in GDP. GDP bottoms out in the fourth quarter at 0.19% below the baseline. When we exogenize the SENSEX-30, GDP shows almost a similar response to a positive overnight call money rate shock. At the beginning of the second year, only 25% of the effects of monetary policy tightening pass through the asset prices. After blocking off the channel, the accumulated response of GDP is reduced by only 24% in twelve quarters. These results suggest that the asset price channel is not important in the transmission of monetary shocks to the real sector in India. 4.3. Exchange rate channel The importance of the exchange rate channel in the transmission of monetary shocks depends on the nature of the exchange rate regime and the degree of openness of the economy. In order to examine the exchange rate channel, we divide the response of the real sector to an interest rate shock into two steps. The first step describes the reaction of the exchange
Fig. 11. Asset price channel: impulse responses to a positive overnight call money rate shock.
A. Aleem / Journal of Asian Economics 21 (2010) 186–197
195
Fig. 12. Exchange rate channel: impulse responses to a positive overnight call money rate shock.
rate to an interest rate shock. The reaction of the exchange rate to an interest rate shock depends on the nature of the exchange rate regime. In a free floating exchange rate regime, the exchange rate reacts more pronouncedly to an interest rate shock. However, if the exchange rate is pegged or heavily managed, it will not respond to the interest rate shock. The second step describes the reaction of the economy to variations in the exchange rate. The response of the real sector to variations in the exchange rate depends positively on the degree of openness of an economy. Since 1993, the official exchange rate regime in India has been the market-determined exchange rate regime. However, frequent interventions by the Reserve Bank of India to stabilize the movements of the Indian rupee vis-a`-vis the US dollar show that the de facto exchange rate regime is different from the de jure exchange rate regime and the Indian rupee seems to be pegged to the US dollar.8 According to the IMF classification of de facto exchange rate regimes in 2006, India has a de facto managed floating regime with an unannounced path of exchange rate (International Monetary Fund, 2006). In 2006, the committee on fuller capital account convertibility affirmed the recommendations of the 1997 committee to the Reserve Bank of India, recommending that the Reserve Bank should maintain a monitoring band of 5% around the real effective exchange rate and should intervene as and when the real effective exchange rate moves outside of this band (Reserve Bank of India, 2006). According to these recommendations, the Reserve Bank of India can use its judgment to intervene even within the band to preclude speculative forces and unwarranted volatility. The committee further recommended that the Reserve Bank of India should undertake a periodic review of the real effective exchange rate, which can be changed as warranted by fundamentals. Moreover, ensuring orderly conditions in the foreign exchange market to avoid excessive exchange rate volatility is one of the objectives of Indian monetary policy. Frequent interventions by the Reserve Bank of India to stabilize the exchange rate weaken the exchange rate channel. Moreover, the large size of the Indian domestic market as compared to its total exports or imports also suggests that the exchange rate channel is not important in the transmission of monetary policy shocks to the real sector. In order to examine the exchange rate channel, we focus on the effects of a positive overnight call money rate shock on GDP that passes through the exchange rate. The vector of endogenous variables consists of GDP, prices, overnight call money rate and real effective exchange rate (REER). Y 0t ¼ ½ GDP
Prices
i
REER
Fig. 12 depicts the responses of GDP, prices and real effective exchange rate to positive innovations in the overnight call money rate and real effective exchange rate. In response to an unanticipated increase in the overnight call money rate, the real effective exchange rate appreciates initially and shows a short-lived reaction to a positive overnight call money rate shock. The effects of a positive overnight call money rate shock on the real effective exchange rate almost disappear in the sixth quarter. A weak response of the real effective exchange rate to a positive monetary policy shock weakens the first step in the exchange rate channel. These results also suggest that the Reserve Bank of India tends to stabilize the real effective exchange rate. The right panel of Fig. 12 shows the responses of GDP to overnight call money rate innovations with and without real effective exchange rate exogenized. In the two cases, the GDP responds to positive overnight call money rate innovations almost in a similar way. These results suggest the absence of an exchange rate channel in India. 4.4. Robustness of results The empirical results in the augmented VAR models suggest the importance of the bank lending channel in India. In order to check the robustness of these empirical results, we examined their statistical significance. We estimated 2S.E. confidence intervals after blocking off each channel. Fig. 13 depicts the impulse responses of GDP to a positive overnight call money rate shock within the channel and after blocking off the channel with 2S.E. confidence intervals.
8
Similarly, Reinhart and Rogoff (2002) found that from July 1995 to December 2001, India had a de facto crawling peg to the US dollar.
196
A. Aleem / Journal of Asian Economics 21 (2010) 186–197
Fig. 13. Response of GDP to a positive overnight call money rate shock.
We find that the responses of GDP to a positive overnight call money rate shock are statistically significant in the bank lending channel. However, GDP shows similar responses to a positive monetary policy shock in the asset price and exchange rate channels. These estimates suggest the robustness of our empirical results and validate the importance of the bank lending channel in India. 5. Conclusion The existence of external constraints on monetary policy in emerging economies requires a model specification different from that of developed countries. This paper provides a comprehensive empirical analysis of the monetary transmission mechanism in India. We estimated a series of VAR models to examine three transmission channels of monetary policy. The benchmark VAR model is composed of a vector of endogenous domestic variables and a vector of exogenous foreign variables. We imposed restrictions on the contemporaneous effects of endogenous variables to have an exact identification of the benchmark VAR model. The results of the benchmark VAR model suggest that an unanticipated monetary policy shock has transitory effects on the overnight call money rate. The price-puzzle vanished after the inclusion of the vector of exogenous foreign variables. Prices and GDP decline after an unanticipated positive overnight call money rate shock. Moreover, prices start declining after a decline in GDP. The Indian economy is a bank-based economy. Since 2005, bank credit to the commercial sector has accounted for more than 70% of total domestic credit. The currency to deposit ratio has declined continuously since 1999. These facts suggest that the banks play an important role in financial intermediation and that the non-financial sector lacks alternative sources of funding. Our empirical results support the importance of the bank lending channel in the transmission of monetary policy shocks to the real sector. The lower market capitalization of listed companies in India as compared to developed countries suggests that the capital markets in India are not sufficiently developed. Empirical estimates in the augmented VAR model suggest that the asset price channel is not important in the transmission of monetary policy shocks to the real sector in India. Central banks in emerging economies stabilize exchange rates even though they announce that they do not do so. Massive interventions by the Reserve Bank of India in the foreign exchange market to stabilize the exchange rate weaken the exchange rate channel. The shortlived response of the real effective exchange rate to an unanticipated monetary policy tightening suggests that the exchange rate channel is not important in the transmission of monetary policy shocks in India. This analysis provides some important theoretical and policy implications. First, Indian monetary policy is constrained by the Fed’s monetary policy. Hence, an analysis of Indian monetary policy requires the inclusion of the federal funds rate in the information set of the Reserve Bank of India. A proper model specification, considering the external constraints on monetary policy and controlling for international economic events, reduces the bias. Second, the Reserve Bank of India intervenes massively in the foreign exchange market to stabilize the exchange rate. The Indian rupee seems to be pegged to the US dollar. Hence, a proper comprehension of the monetary transmission mechanism in India requires the analysis not only of the response of GDP, but also of the response of the exchange rate to a monetary policy shock. Third, banks play an important role in financial intermediation in the Indian economy, and their strong representation reflects the lack of alternative sources of funding for the private sector. Appendix A. Data sources Overnight call money rate, bank rate, repo rate, reverse repo rate, bank credit to commercial sector, currency in circulation, time deposits, demand deposits: Reserve Bank of India; and Reuters. Wholesale Price Index, federal funds rate, GDP of the United States and India: International Monetary Fund, International Financial Statistics.
A. Aleem / Journal of Asian Economics 21 (2010) 186–197
197
Sensitive index (BSE-30), spot rates: Reuters; and Eurostat. Real effective exchange rate: The Bank for International Settlements. Prime lending rate: OECD. World commodity price index: The economist. Market capitalization of listed companies: World Bank, World Development Indicators. References Ahmed, N., Hastam, S., Asif, A. I., & Yasir, M. A. (2005). Transmission mechanism of monetary policy in Pakistan. SBP research bulletin, Vol. 1, No. 1, State Bank of Pakistan. Al-Mashat, R. (2003). Monetary policy transmission in India: Selected issues and statistical appendix. Country report no. 03/261. International Monetary Fund. Barran, F., Coudert, V., & Mojon, B. (1996). The transmission of monetary policy in European countries. Working paper no. 96-03. Paris: CEPII. Bernanke, B., & Blinder, A. (1988). Credit, money and aggregated demand. The American economic review, papers and proceedings of the one-hundredth annual meeting of the American economic association, Vol. 78 (pp. 435–439). Bernanke, B., & Blinder, A. (1992). The Federal funds rate and the channels of monetary transmission. The American Economic Review, 82, 901–921. Calvo, G., & Reinhart, C. (2000). Fear of floating. NBER working paper no. 7993. Cambridge: National Bureau of Economic Research. Christiano, L. J., Eichenbaum, M., & Evans, C. (1998). Monetary policy shocks: What have we learned and to what end? NBER working paper no. 6400. Cambridge: National Bureau of Economic Research. Disyatat, P., & Vongsinsirikul, P. (2003). Monetary policy and the transmission mechanism in Thailand. Journal of Asian economics, 4, 389–418. Favero, C. A. (2004). Applied macroeconometrics. Oxford University Press. pp. 162–211. International Monetary Fund. (2006, July). De facto classification of exchange rate regimes and monetary policy framework. IMF. Available on http://www.imf.org/ external/np/mfd/er/2006/eng/0706.htm. Kannan, R., Sanyal, S., & Bhoi, B. (2006, Winter). Monetary conditions index for India. Reserve Bank of India occasional papers 27, no. 3. Levy-Yeyati, E., & Sturzenegger, F. (2005). Classifying exchange rate regimes: Deeds vs. words. European Economic Review, 49, 1603–1635. Morsink, J., & Bayoumi, T. (2001). A peek inside the black box: The monetary transmission mechanism in Japan. IMF staff papers, Vol. 48, International Monetary Fund. pp. 22–57. Pandit, B. L., Mittal, A., Roy, M., & Ghosh, S. (2006). Transmission of monetary policy and the bank lending channel: Analysis and evidence for India. Study number 25, Mumbai: Development Research Group, Department of Economic Analysis And Policy, Reserve Bank of India. Peersman, G., & Smets, F. (2001). The monetary transmission mechanism in the euro area: More evidence from VAR analysis. Working paper no. 91. Germany: European Central Bank. Prasad, A., & Ghosh, S. (2005). Monetary policy and corporate behaviour in India. IMF working paper no. 05/25. International Monetary Fund. Ramaswamy, R., & Sloek, T. (1997). The real effects of monetary policy in the European Union: What are the differences? IMF working paper 97/160. Reinhart, C., & Rogoff, K. (2002). Modern history of exchange rate arrangements: A reinterpretation. NBER working paper no. 8963. Cambridge: National Bureau of Economic Research. Reserve Bank of India. (2003, March). Money, credit and prices. Report on currency and finance. Mumbai: Reserve Bank of India. Reserve Bank of India. (2004, December). Monetary transmission mechanism. Report on currency and finance 2003-04 (pp. 171–193). Mumbai: Reserve Bank of India. Reserve Bank of India. (2006, July). Chapter 5: Interaction of monetary policy and exchange rate policy. Report of the committee on fuller capital account convertibility. Sims, C. (1992). Interpreting the macro-economic time series facts: The effects of monetary policy. European Economic Review, 36, 975–1011. Sims, C., Stock, J. H., & Watson, M. W. (1990). Inference in linear time series models with some unit roots. Econometrica, 58, 113–144. Singh, K., & Kalirajan, K. (2007). Monetary transmission in post-reform India: An evaluation. Journal of Asia pacific economy, 12, 158–187. Suzuki, T. (2004). Credit channel of monetary policy in Japan: Resolving the supply versus demand puzzle. Applied Economics, 36, 2385–2396. Virmani, V. (2004). Operationalizing Taylor-type rules for the Indian economy: Issues and some results (1992Q–2001Q1). IIMA working paper no. 2004-07-04.