تجزیه و تحلیل اثرات شوک سیاست پولی ایالات متحده در کشورهای دلارریزه شده

Identifying monetary policy shocks is difficult. Therefore, instead of trying to do this perfectly, this paper exploits a natural setting that reduces the consequences of shock misidentification. It does so by basing conclusions upon the responses of variables in three dollarized countries (Ecuador, El Salvador, and Panama). They import US monetary policy just as genuine US states do, but have the advantage that non-monetary US shocks are not imported perfectly. Consequently, this setting reduces the effects of any mistakenly included non-monetary US shocks, while leaving the effects of true monetary shocks unaffected. Results suggest that prices fall after monetary contractions; output does not show a clear response

Since monetary policy is typically not executed in an erratic fashion, identifying random disturbances to monetary policy (the so-called “monetary policy shocks”) is difficult. For this reason, the present paper has a different focus than the standard VAR exercise. Instead of trying to find the perfect shock identification scheme, this paper asks: if shock identification is so difficult, cannot we find a natural setting that reduces the consequences of the almost inevitable misidentification of monetary shocks? The natural setting I exploit is the existence of dollarized countries. They import US monetary policy, but, as I will argue below, they are not perfectly integrated with the US economy. Consequently, non-monetary US shocks do not survive the transmission process to these client economies undamaged, which makes this paper's findings less prone to monetary policy shock misidentification. The fact that shock identification is difficult, might explain the presence of some ongoing debates in the structural VAR literature. Next to the fact that there is no consensus on the effects of monetary shocks on output, many studies find that prices increase after a monetary contraction, which goes against the predictions of most standard macroeconomic theories (such as the New Keynesian one). Even though this price response can be rationalized through the working capital channel, 1 it is generally referred to as “the price puzzle”. The present paper tries to shed light on the question whether economic theory should take this price puzzle seriously (which may call for incorporation of the cost channel into standard macroeconomic models2), or whether it is just an artifact of shock misidentification. It does so by using output and price data from dollarized countries, all located in Latin America. By unilaterally adopting the US dollar, these countries have established a so-called “informal monetary union” with the US. From a monetary perspective, these client countries are therefore not that different from genuine US states: they use the US dollar as legal tender (just like, say, Idaho does) and have no possibility to deviate from US monetary policy, as there is no local currency to de- or revalue. Consequently, these countries rapidly import US monetary shocks (primarily via the financial channel; see Canova, 2005), while there are no exchange rate considerations at play. The fact that the monetary union is only “informal” (thereby contrasting with formal monetary unions, such as the euro zone), does not matter in this respect. Taking this geographical detour brings at least two advantages. Firstly, the resulting econometric restrictions enable one to analyze the effects of US monetary shocks in the client economies, without imposing inertial or sign restrictions on the variables of interest. Secondly, basing conclusions upon the responses of variables in dollarized countries makes this paper's findings less prone to the major concern any structural VAR exercise has to face: misidentification of the US monetary shock. This is the case because the dollarized countries that are going to be considered (Ecuador, El Salvador, and Panama) are only imperfectly integrated with the US economy. In particular, the economies of Ecuador and El Salvador are only moderately open in terms of trade-to-GDP ratios3 and to the extent that the dollarized economies do trade internationally, most of it takes place with other countries than the US4 Consequently, non-monetary US shocks can be expected to produce only rather limited output – and price fluctuations in these countries – especially at short horizons.5 Lindenberg and Westermann (2010) investigated this issue empirically and they indeed find that Latin America does not share its business cycle with the US.6 This suggests that these cycles are not driven by the same shocks, or that the shocks are only transmitted with a delay. In line with this, Canova (2005) even finds that non-monetary US shocks do not tend to produce significant output or price fluctuations in Latin America at all. All this suggests that even if the identified “monetary policy shock” includes some non-monetary US components, the consequences of this mistake are contained in the dollarized countries, as the transmission of these non-monetary US disturbances to Latin America is not instantaneous and perfect. This makes the approach work a bit like an ideal filter, as it reduces (or at the very minimum: delays) the effects of any mistakenly included non-monetary US shocks on client country variables. When one analyzes the effects of contractionary US monetary shocks through dollarized economies, prices in all client countries fall on impact—so the price puzzle disappears. Quantitatively, prices in dollarized economies seem to have been pretty flexible over the sample period. Output does not show a clear response, so monetary neutrality cannot be rejected.

This paper has presented an alternative way of analyzing the effects of US monetary policy shocks. The approach is akin to the exploitation of a natural experiment—formed by the fact that there exist countries that have established an informal monetary union with the US (thereby importing US monetary shocks), while being only imperfectly integrated with the US economy (as a result of which non-monetary US shocks are not imported fully and instantaneously).25 Consequently, conclusions based upon the responses of variables in these dollarized economies, are less vulnerable to misidentification of the monetary shock in the US block of the system. In this sense, the natural setting that this paper exploits thus works a bit like a convenient filter that removes any mistakenly included non-monetary elements of the shock. Results suggest that the US “price puzzle” is due to monetary policy shock misidentification, as prices in all dollarized economies fall immediately after a contractionary monetary shock. This finding supports the key identifying assumption that is typically made in studies employing sign restrictions. More generally, obtaining a better insight into the sign of the price response to monetary shocks is important as this determines whether monetary policy is able to stabilize the economy or not. After all: if prices would go up after a monetary contraction, the conventional policy of increasing the interest rate to fight inflation would be like throwing oil on fire. Quantitatively, results indicate that the price effects of monetary shocks are large and show up quickly. This suggests that prices were relatively flexible in the dollarized economies over the sample period. Consistent with this, monetary shocks do not seem to have had a clear effect on output in the considered countries. Investigating to what extent these findings carry over to the US economy itself, could be a fruitful topic for further research: after all, the obvious cost of this paper's approach is that no attempt is made to identify the US monetary shock and inference does not take place on US variables.26Gobbi and Willems (2011) try to come up to this challenge by identifying US monetary shocks through sign restrictions on prices in the dollarized countries only (leaving the responses of US output and prices unrestricted). In line with this paper's results for the dollarized countries, they find that US prices fall after a monetary contraction, while US real GDP fails to show a clear response in their study as well. Finally, it might also be interesting to estimate a panel-VAR that features US variables along with variables of all dollarized countries simultaneously. This enables one to distinguish between country-specific and common shocks, which should be helpful in identifying US monetary shocks (as the latter are common to all countries using the US dollar). Once one has identified the US monetary shock in such a setup, one can also analyze how the various real exchange rates respond to a monetary contraction which might produce interesting insights along that dimension.