Note: This posting is based on remarks delivered at a High-Level Seminar on the Liberalization and Management of Capital Flows at the International Monetary Fund, April 18, 2013. Among the other speakers were the central bank governors of Brazil, Israel, and the Philippines.
Members of the International Monetary Fund (IMF) and the staff and management of the Fund should be congratulated for taking up the complex and vexed topic of international capital flows and reaching consensus on an “institutional view” to help guide their discussions and advice.1 They have taken an important step toward an improved understanding of and dialogue about the policy issues. For some time, I have urged that these issues be addressed, and I am gratified with the results.2
The test will be implementation, however. Can the institutional view be made operational? Will the framework help identify negative effects—known in IMF-speak as negative spillovers—on other countries from policies designed to affect capital flows and counsel appropriate and effective policy adjustments? I am skeptical.
We do not have a robust, well-developed, widely accepted framework to analyze the influence of policies and conditions in recipient and source countries on capital flows. My skepticism is based on three considerations: measurement issues, constraints on economic analysis of a general equilibrium system, and challenges in arriving at robust policy advice to benefit the global economy and financial system.
First, with respect to measurement issues, it is natural to hypothesize that recorded capital flows from country A (let’s say the United States) to country B (let’s say Brazil) can be modeled simply as a function of economic variables in the United States (the push variables) and in Brazil (the pull variables). I choose Brazil for expository purposes not only because some Brazilian officials have been among the most vociferous critics of the negative spillover effects of US quantitative easing, and because Governor Alexandre Tombini participated in the seminar at which I delivered an earlier version of these remarks, but also because Brazilian economists, including at the Central Bank of Brazil, have endeavored to buttress that case.3
The principal measurement problem with testing this hypothesis is that we do not measure capital flows contemporaneously with the factors affecting those flows. In particular, we do not capture ex ante price and quantity demand or supply relationships. We take our readings from the ex post intersections of supply and demand and an average of the prices (interest rates and exchange rates), and we record as a flow the associated cumulative net change in stocks of various categories of assets and liabilities in the interval since our previous reading. True, the sum of those net changes, without regard to sign, across all countries participating in the global economy, is a large and growing number. In 2012, for the United States alone, the figure was $6.7 trillion.4 But the resulting figure is a net figure that is fundamentally flawed for use in economic analysis, masking much larger gross flows, and only observed ex post.
Consider the United States alone. In 2012, the US current account deficit, equal to the net financial inflow to the United States, was $475 billion. The actual, recorded net financial inflow was $400 billion.5 One category of net inflows was net foreign purchases of $548 billion in US long-term securities. That net inflow was only partially offset by $28 billion in US net purchases of long-term foreign securities. Were these later net purchases, which were small, responding to US push factors or foreign pull factors? One cannot tell from the ex post, net data. Total US transactions in foreign, long-term securities over the 12 months of 2012 were $14 trillion, more than 500 times the net figure. Were the gross sales pushed out of the United States, while the almost equal amount of gross purchases pulled into the county? We do not have a clue!
My second concern, with respect to analytical constraints, is that global capital flows at the micro level of pushes and pulls have to satisfy two ex post, equilibrium conditions. Each country’s cross-border transactions, in principal, sum to zero. Similarly for the world as a whole, each country’s import of goods, services, or capital is another country’s export.6 More important, in satisfying these equilibrium conditions, a country’s international financial accounts are affected by conditions in its real economy. These conditions are in turn influenced by those global capital flows, as well as by the country’s own policies and other exogenous factors. Many prices and quantities of assets, as well as of goods and services, are adjusting at the same time.
Thus, capital flows are embedded in a complex, poorly understood, general equilibrium system. If one category records ex post an increase in net inflows, some other category must record a decrease in net inflows or increase in net outflows—including changes in official accounts. Otherwise, the current account (the counterpart to overall net capital flows) must have moved toward deficit or into greater deficit.
Let me return to the United States and Brazil to illustrate this point. Assume that the Federal Reserve embarks on a policy of quantitative easing (QE). Will this policy produce an increase in net capital outflows from (reduced inflows to) the United States? One indicator is whether the US current account deficit narrows. IMF and Federal Reserve macroeconomic models, which may be flawed for the reasons I have described, suggest that the aggregate net effect of easier US monetary policy on the US current account balance and net exports is effectively zero. (Again, the US current account balance is the same as the US net capital flow, with the sign reversed.) The reason is that, as an empirical matter, the sign and size of the effect on US current account via any dollar depreciation associated with the quantitative easing is offset by the opposite sign and commensurate size of the effect on that same account via any resulting boost in US domestic demand.
The data show that, over 2011 and 2012, the US current account deficit has been unchanged and the dollar’s real effective exchange rate has fluctuated over a narrow range of less than 7 percentage points, with no trend toward depreciation. 7 These data are only suggestive, but they are consistent with the view that the aggregate net effect of US quantitative easing on the rest of the world has been trivial.
What about capital flows to Brazil? Accept, for the moment, that the net effect of US quantitative easing on the US current account balance is zero. This does not mean that the net effect on the current account position of every US partner country is also zero. The net effect might reasonably be positive for Canada and Mexico and possibly negative for Brazil. Exchange rate effects may be stronger for some countries than for others, and the reverse for aggregate demand effects on trade. At the macro level, any negative effect on Brazil’s current account position would have had to be accommodated by an appreciation of the real or by faster growth in Brazilian domestic demand. We observe that, in 2011 and 2012, the real has depreciated, Brazilian growth has slowed, and its current account deficit has widened somewhat. It is difficult to conclude that these developments had much to do with US quantitative easing.
My conclusion is that it is a challenge to tell a coherent story about push and pull policy factors affecting international capital flows—one that policymakers can rely upon to guide their policies.
Partial-equilibrium analysis is insufficient to establish a causal chain running from quantitative easing in the United States to an increase in net capital flows to Brazil. Partial-equilibrium analysis that focuses on interest rates alone, such as the spread between long-term US Treasury rates and the rate on three-month Treasury bills, not only ignores the effects of faster growth of US domestic demand but also abstracts from other potential effects on Brazil external accounts. Those effects include increases in Brazilian foreign exchange reserves (recycling private capital inflows); adjustments in the Brazilian real and financial economy, induced by US easing or associated with contemporaneous Brazilian policy actions; and adjustments induced by exogenous policy actions in other countries.
This brings me to my third and final consideration: the challenge in arriving at robust policy advice for the good of the global economy and financial system. Continuing with the example of US quantitative easing, what is the negative spillover from US quantitative easing that the IMF should bring to the attention of the US authorities? What does the IMF institutional view suggest that the US authorities should do about it?
Capital flows are measured ex post, and the measurements may not capture responses to changes in pushes and pulls or other influences. Moreover, capital movements are essentially endogenous within a general equilibrium system, responding to many other macroeconomic policy and environmental influences. Thus, it is difficult to identify capital flows to one country with policy actions in any other country.
Assume, nevertheless, that we could establish a connection between US quantitative easing and negative or positive economic aggregate macroeconomic effects on the rest of the world that are evenly spread across US partner countries. It is likely that any positive effects would be welcome in countries experiencing subpar growth (Europe) and not welcome in countries that are overheating (some emerging market economies, at least until recently), and vice versa. In other words, it would be very difficult to generalize about the negative effects for the world as a whole and frame the advice to US authorities accordingly
More broadly, it has long been demonstrated that conventional monetary policy actions in the United States and other systemically important countries affect macroeconomic conditions in other countries via sympathetic movements in interest rates in those countries, shifts in risk aversion, or other indirect channels.8 These effects need not show up in changes in current account balances or net capital flows. Is quantitative easing any different? Should country A be expected not to address full employment and price stability simply because its actions may hurt other countries? That is a difficult case to make. Broad macroeconomic influences should not be conflated with policies affecting capital flows per se. Rather they should be addressed in the context of the surveillance of countries’ overall policies.
In summary, the new IMF “institutional view” on the management of capital flows and the welcome associated research, unfortunately, have not provided us with a robust framework for analyzing the push and pull factors affecting international capital flows. This is not to say that we should stop trying to develop a better operational consensus on these issues. Nor is it to say that policy actions in the United States, or any other important economy—including Brazil, Israel, and the Philippines—have no effect on capital flows or on other economies, and therefore can be, or should be, ignored. They are an appropriate matter for global discussion. But discussions with the IMF and among its members must account for all relevant factors, and be well-grounded empirically and analytically. Progress has been made, but we are not there yet.
1. IMF (International Monetary Fund). 2012. The Liberalization and Management of Capital Flows: An Institutional View (November 24). Washington: International Monetary Fund.
2. Edwin M. Truman. 2010. Dealing with Volatile Capital Flows: The Case for Collective Action. Peterson Institute for International Economics Real Time Economic Issues Watch, May 25, 2010. Available at www.piie.com. Edwin M. Truman. 2011. Taking Stock of Volatile Capital Flows: Progress Despite the Noise. Peterson Institute for International Economics Real Time Economic Issues Watch, April 20, 2011. Available at www.piie.com.
3. See Joao Barata R. B. Barroso, Luiz A. Pereira da Silva, and Adriana Soares Sales. 2013. Quantitative Easing and Related Capital Flows into Brazil: Measuring its Effects and Transmission Channels through a Rigorous Counterfactual Evaluation. Banco Central do Brasil Working Paper presented at a Technical Seminar on Spillover Methodologies, April 21, 2013.
4. This figure is derived from summing the changes in the individual lines in the US international investment position published by the Bureau of Economic Analysis of the US Department of Commerce.
5. The difference is net errors and omissions in the accounts.
6. In addition, in market for foreign exchanges, which we can analyze using microeconomic tools, supply must equal demand, accounting for changes in inventories. This is a big market. In 2010, annual global foreign exchange turnover was on the order of $1,000 trillion, and capital flows need not involve foreign exchange market transactions.
7. I am using the Federal Reserve Board staff’s broad index.
8. This is my interpretation of the paper by Barroso et al. cited in footnote 3 above. It demonstrates that Federal Reserve quantitative easing has been associated with effects on the Brazilian economy including in measured capital flows, but it does not demonstrate that those effects are unique to the quantitative easing, are directly associated with that policy, or are substantially different from the effects of conventional monetary policy actions by the Federal Reserve.