The Indian Rupee in International Trade

floating-rupee

Exports and remittances in foreign currencies by non-resident Indians (NRIs) have historically constituted a large majority of foreign currency reserves of India. Post-colonial Fabian socialist policies – which mostly closed off the Indian market to investment by foreigners – also restricted the Indian currency, the rupee, to a fixed (1947-1977), a basket peg (1978-1992), and a dual exchange rate system (1992-1993), in this order of historical evolution, until 1993.

Beginning in 1991, economic reforms in response to the balance of payments crisis saw the Indian exchange rate regime also evolve into an unified exchange rate system by March 1993. The unification of the exchange rate of the Indian rupee was an important step towards full convertibility on all current account transactions which was finally achieved in August 1994 when India accepted obligations under Article VIII of the Articles of Agreement of the International Monetary Fund (IMF). Upon extensive consultations with experts and market participants, the Reserve Bank of India (RBI) gradually implemented wide-ranging reforms in the second half of the 1990s to remove market distortions and deepen the foreign exchange market.

The focus of the reforms has been to move away from the micromanagement of foreign exchange transactions to the macro-management of foreign exchange flows. Because the RBI at times actively intervenes to “correct” the exchange rate, India can be considered to be a managed float regime rather than an independently floating exchange rate system. To become an independently floating currency, the controls on the capital account must fully be lifted just as they were lifted on the current account: India’s capital account, which consists of foreign direct investment (FDI) and foreign institutional investment (FII), must become fully convertible.

India is currently the seventh largest economy by nominal gross domestic product (GDP) and on a purchasing power parity (PPP) basis it is the third largest economy in the world. The trilemma does not hold for large economies. India is growing rapidly and is by no means a small economy but even if the trilemma were to apply to India it can retain control over its monetary policy and have open capital markets and move to an independently floating exchange rate regime.

Most importantly the growing up of India from the status of a lower middle income country to a wealthy nation will depend upon India emulating the United States for the growth of its domestic market and Germany (despite the strong euro) for the growth of its export market to be able to weather the vicissitudes of the global economy for sustaining – by mitigating the risks of free financial flows across borders – an independently floating currency.

While removing controls on the capital account, India can improve its international economic standing further by gradually beginning to use its own currency in global trade by entering into currency swap arrangements with its major trading partners – India must begin to pay for its imports in rupees rather than in dollars or in other major currencies. The internationalization of the Indian rupee, including for investment in rupees by Indian firms in foreign markets, will make the rupee a reserve currency in international trade thus minimizing the need for maintaining large foreign exchange reserves.

The Other Side of Central Bank Independence

cbi

Former Reserve Bank of India (RBI) governor Raghuram Rajan warned that low interest rate policies by central banks could distort the economy and may be difficult to exit out of. Three common market distortions are inflated stock prices, lower bond yields and higher bond prices for both government and corporate bonds, and downward pressure on currencies.

Difficulty in exiting a low interest rate environment, especially when inflation is globally low because of sluggish growth, arises primarily because central banks do not see the need to exit as inflation is lower than their targets and secondarily because monetary authorities fear that rising interest rates can put downward pressure on growth and drive the economy back into a recession.

The reality is that the markets are constantly negotiating with the central bank for ever lower rates to be incentivized to raise real investment. In fact, instead of raising real investment, public corporations are boosting their price-to-earnings ratios by raising capital cheaply through bond issuance and using that money to buy back stock to artificially elevate their stock prices. This market distortion seriously undermines the credibility of the argument that ever lower rates are needed for real investment to rise.

Further, sluggish growth becomes another reason for holding back real investment because corporations do not see earning potential in an environment of tepid demand and slack in production capacity – the slack reduces the need for new capital investment and slows the pace of retooling existing capital.

Moreover, if consumers become averse to new debt having entered the recession with higher debt-to-income ratios, coming out of the recession, despite higher real money balances, they may be unwilling to spend as much but instead prefer precautionary saving and reduction of their debt burden.

This spiral of low investment, low consumption and low growth, however low the central bank interest rate may be, leads to “pushing on a string” as famously quipped by John Maynard Keynes or impotent monetary policy until the government engages in fiscal expansion to increase government investment for job creation which can raise wages, consumption and end disinflation or deflation as the case may be. Otherwise, economic recovery would be slow, prolonged and lackluster as has been happening in the developed world. Therefore, low interest rates alone are not the answer to recovering from deep recessions though monetary policy could be the correct response for short, shallow dips in growth.

Stanley Fischer, the Vice Chairman of the Federal Reserve, recently gave three reasons for real investment undershooting Fed’s projections: hightened uncertainty, low capital intensity and developments outside the US. If, in particular, the US economy’s structure has changed to one of low capital intensity, then why continue the crisis era low interest rate policy but to encourage the unwelcome market distortions arising from cheap money especially when, unlike German and Italian banks, US banks have been well capitalized after the financial crisis of 2007-2008?

In India, Raghuram Rajan was politically criticized for not lowering interest rates more than he had. Unlike in the US, non-performing assets (NPAs) are a problem in China and India as they are in Italy and this is partially why monetary policy is not being effectively transmitted for raising real investment. Yet the political pressure on the central bank – contrary to the cornerstone economic principle of independent central banking – increases to cut rates further. The real sources of this pressure are the large corporations lobbying their governments, just as they also do in the United States, for low interest rates.

While central banks have become statutorily independent from their governments, given their mandates to come to the rescue of financial institutions they cannot be independent from the influence of the markets. This is the other side of central bank independence. To be truly disinterested in the affairs of both the markets and the governments, central banks should not cater to the bottomless low rate hunger of the markets because of known market distortions of needlessly prolonged expansionary policies. The central banks must work with their governments in dynamic interdependence so that there is a sound fiscal-monetary mix in stabilization policy as a part of a robust and politically independent macroeconomic policy framework.

Rajan has cautioned wisely once again and it would be sapient to heed such words.

How Can India Grow More? Comparing the Economies of the United States, China and India

indiaeconomy

India is the seventh largest economy by nominal gross domestic product (GDP) and the third largest economy in the world on a purchasing power parity (PPP) GDP basis. After watching the miracle of near-continuous strong growth unfold in China over the last 30 years catapulting China to the position of the world’s largest economy on a PPP-basis, the expectation is for India to follow, though in some quarters there is disappointment that India is significantly lagging behind China and that it may be difficult for India to catch-up.

China is an industrialized country now but is yearning to be categorized as a market economy. There is still substantial government intervention in the Chinese economy and the country continues to be a state managed by a single political party. China does not intend to be a pluralistic democracy and has been unequivocal in stating that it wants to remain a monolithic political system, run by the Chinese communist party, combined with a gradually evolving free market economic system. It may always be the case that the Chinese economy could continue to have significant government intervention, with large state owned enterprises co-existing with private corporations. China is yet to fully develop its financial markets to become as broad and deep as that of the United States. With its Fabian socialist past, such has also been the case with India’s economy but the country has developed a robust private sector since market reforms in 1991.

The key distinction between China and India is that India has been a full-fledged democracy and a mixed economy since its independence from Great Britain in 1947. It is yet to be seen if the democratic politics of India are holding back a China-type economic development in India, its multiparty politics sapping the government of any political will to provide the key determinants of growth such as broad and deep financial markets to promote real investment and infrastructure (including, besides road, rail and air transportation infrastructure, energy, water and sanitation infrastructure) for India to develop its large domestic market.

It raises the issue of sequencing of economics and politics: can China evolve politically into a Hong Kong or Singapore after first economically developing under an authoritarian political system for about three decades? Do democracies need more time and political will to achieve the same? After all, US and European infrastructure development were either in preparation for war or as a consequence of war (World Wars I and II). This does not mean that India ought not to be a democracy to achieve the economic development it deserves.

India has a strong democratic tradition of politics and robust political institutions to support democracy. It needs similar economic institutions to make the transition to a fully open market economy with a fully convertible currency in the global currency markets with no controls on either the current account or the capital account. To achieve it, India needs superior first-world infrastructure which can take about three decades to build and a strong manufacturing base to buttress the agricultural and services sectors.

In structure, thanks to the economic reforms of 1991, India is an economy, ahead of China, already poised to emulate the American economy whose domestic market, without understating the importance of exports, contributes a lionshare of its GDP. The reasons behind the success of the U.S economy are infrastructure and innovation. Despite the size of the US market being a fourth of the Indian or Chinese market in population, cycles of product and process innovation continuously bring back consumers to the marketplace for newer versions of products and services. India and China should aspire to the same. Superior infrastructure and innovation can then help India and China substantially leverage their highly populated markets to grow in a manner that is in sync with the premise of sustainable development – so that growth is inclusive, broad-based and environmentally friendly. On this front, India is a laggard because it is an economy where the gap between the rich and the poor is highly unequal.

India is on the correct path. It only needs to institutionalize its growth strategies so that democratic politics do not make the economic development landscape uncertain. “Make in India” will increase the size of the manufacturing sector if the government provides market access to foreign corporations provided they manufacture in India, creating Indian jobs. Besides exports, this will also develop the domestic Indian market for goods and services produced in India. “Startup India” will change the environment for innovation besides of course the need for research and development by Indian and foreign corporations to take place in India to feed the product cycle to increase domestic consumption.

Broad and deep financial markets, infrastructure and innovation can significantly raise Indian economic growth for the country to develop on a sustainable path to make India perhaps the largest democracy and economy by 2050.

Machine Learning and Economic Analysis

machine_learning_alogrithms

One definition of machine learning is the study and construction of algorithms that learn from and make predictions on data. Rather than programmed to do certain tasks, machine learning algorithms build models from inputs to make data-driven predictions or decisions.

Forecasting or making predictions is also the bread and butter of the science of econometrics which takes measured economic data and builds models of the relationships between the various data elements to explain how a dependent variable is influenced by changes to one or more independent variables. The standard approach in econometrics is to fit data to a straight line (also known as a linear regression) and use the equation of the straight line to obtain the quantitative relationship between the dependent variable and the one or more independent variables.

The primary difference between econometrics and machine learning is that the former builds models to test hypotheses and economic theories while the latter begins with the data and builds models that identify patterns in the data. The very large size of the available datasets that is commonplace in machine learning is increasingly becoming unwieldy for traditional econometric analysis to handle. With the coming of Big Data techniques of manipulating very large datasets, it is now increasingly attractive for economists to sieve through a large number of explanatory variables to find interesting relationships between data elements that were difficult to unearth using standard econometric techniques. Therefore, machine learning, particularly on Big Data, promises to have enormous value for economists.

Broadly, machine learning can be categorized into supervised and unsupervised learning. While supervised learning typically entails using a set of “features” or “covariates” (x’s) to predict an outcome (y), unsupervised learning consists of clustering similar data together and does not involve the prediction of an outcome. In economic forecasting, we are interested in prediction and, therefore, we typically deal with supervised learning.

The goal of prediction is always to get good out-of-sample forecasts. Many a time good in-sample predictions do not result in good out-of-sample forecasts. In machine learning, the problem of near perfect in-sample fits not also producing good out-of-sample forecasts is known as the “overfitting problem.” Simpler models tend to work better for out-of-sample forecasts. Economists tend to prefer simpler models for the same reason, but have not been as explicit about quantifying complexity costs. In machine learning various ways to penalize models for excessive complexity is known as “regularization.” Regularization and systematic model selection will become a standard part of empirical practice in economics as economists more frequently encounter datasets with many covariates, and also as economists see the advantages of being systematic about model selection.

It is conventional to divide the data into separate sets for the purpose of training, testing, and validation. Training data to estimate a model, the validation data to choose a model, and the testing data to evaluate how well the chosen model performs (often validation and testing sets are combined). One common feature of many machine learning methods is that they use cross-validation to select model complexity; that is, they repeatedly estimate a model on part of the data and then test it on another part, and they find the “complexity penalty term” that fits the data best in terms of mean-squared error of the prediction (the squared difference between the model prediction and the actual outcome).

The test-train cycle and cross-validation are very commonly used in machine learning and should be used much more in economics particularly when working with large datasets. For many years, economists have reported in-sample goodness-of-fit measures using the excuse that economists had small datasets. But now that larger datasets have become available, there is no reason not to use separate training and testing sets. Cross-validation also turns out to be a very useful technique, particularly when working with reasonably large data. It is also a much more realistic measure of prediction performance than measures commonly used in economics.

Machine learning prediction models are built on a premise that is fundamentally at odds with a lot of social science work on causal inference. The foundation of supervised machine learning methods is that model selection (cross-validation) is carried out to optimize goodness of fit on a test sample. A model is good if and only if it predicts well. Yet, a cornerstone of introductory econometrics is that prediction is not causal inference. This type of model has not received almost any attention in machine learning.

Currently, in these early days, most machine learning methods are “off-the-shelf” algorithms which are implemented to process large to very large datasets primarily for the purpose of patttern recognition in data and prediction. In particular, machine learning algorithms for prediction carry greater predictive power than standard econometric models. These methods, as the techniques advance and mature, will be modified and tailored for the needs of social scientists who are primarily interested in conducting inference about causal effects and estimating the impact of counterfactual policies.