AN EMPIRICAL MODEL OF FED SWAP LINE SELECTION

If the claim that the Fed acted as an ILLR in order to protect the US financial system is correct, then we would expect a positive statistical relationship between foreign exposure of major banks and money market funds and swap line selection. Major commercial financial institutions are a well-organized, well-financed political lobby. It is highly likely they recognized the substantial threat that the global dollar shortage posed to their profits and, in some cases, their very survival. I anticipate that these institutions pressed US economic officials to provide emergency liquidity to foreign jurisdictions where their portfolios were most at risk. However, as I argued in chapter 5, it is overly simplistic to portray the Fed policymakers as mere puppets of private financial actors. Policymakers are individuals, operating inside state institutions, with their own interests over policy. One of the Fed’s primary roles is to maintain the stability of the US financial system and contain systemic risk in financial markets. Thus, I anticipate that policymakers should be most sensitive to financial institutions’ pleas for protection when the health of the entire system is threatened. As I have already explained, in the fall of 2008, the threats facing the US financial system were existential. In short, during the Great Panic, the interests of major private financial institutions and the Fed’s interest in upholding its public mandate were in close alignment. If my argument is correct, the Fed should have been most likely to deploy its resources on behalf of foreign economies where systemically important US banks and money market funds were most exposed.

In order to test this argument, I estimate an empirical model of the Federal Reserve’s swap line selection against a sample of all countries that had accepted the IMF’s Article VIII by 2008. This has historically been the Fed’s "red line” for swap line selection. Thus, it is unreasonable to include nonsignatories in the sample since there is essentially no way the Fed would have provided liquidity lines to these countries.[1] The key explanatory variable in the model accounts for US financial system exposure in 2007, just before the crisis erupted. It combines cross-national variation in the foreign claims of SIFIs and US residents’ holdings of foreign debt securities into one measure. As discussed in chapter 5, SIFIs lie at the core of domestic banking systems. As such, they are viewed as having more "systemic importance”—defined as "the damage a bank’s failure inflicts upon the rest of the system”—than smaller banks.[2] Foreign debt securities data include commercial paper and negotiable certificates of deposits. These account for US money market fund exposure to foreign jurisdictions. In sum, the key covariate in the model accounts for crossnational variation in the exposure of the US financial system as a percentage of total SIFI foreign claims and residents’ foreign securities holdings. A higher percentage equates to greater US financial system exposure. This variable is discussed in detail in the appendix.[3] As exposure to a foreign jurisdiction increases, I anticipate that the likelihood of a Fed swap line should increase.

Additionally, I follow Lawrence Broz in including several control variables drawn from two federal reports on the swap program.[4] These reports point to four factors that contributed to the Fed’s selection

Table 7.3 FEDERAL RESERVE SWAP LINE REGRESSION RESULTS,2008

Model 1

Model 2

Intercept

-3.513**

* (1.290)

-3.407

(2.3297)

Share ofworld GDP (%)

-2.048

(1.379)

-5.647*

(1.8235)

Inflation

-0.324

(0.340)

-0.259

(0.4404)

Share of US trade (%)

-1.054+

(0.573)

-2.289*

(1.0772)

Global financial center

0.474

(1.384)

-3.491

(8.0470)

US financial system exposure (%)

10.627*

(4.497)

19.099*

(7.4864)

Dollar liquidity shortage

0.009

(0.0468)

N

63

33

AIC

28.947

26.415

Heteroskedastic and Autocorrelation Consistent (HAC) standard errors in “( )”: t p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001.

process: the economic size of the partner, its record of sound economic management, its importance as a major US trading partner, and whether it was a major financial center. To account for these factors, I control for a country’s share of world GDP, the level of inflation (as a proxy for sound economic management), a partner’s share of US trade, and whether or not the partner is home to a global financial center. Finally, in a second model, I also include a measure of dollar scarcity.[5] This controls for the extent of the “dollar shortage” facing foreign jurisdictions. As I explained earlier, the crisis resulted in the seizure of global dollar-funding markets because of fears of counterparty risk. However, some jurisdictions were hit harder by the dollar “shortage” than others. A lower score on this measure indicates a more intense shortage of dollar liquidity. Allen and Moessner find that increased dollar scarcity is associated with a higher probability of receiving a swap line.[6]

Results for two logistic regression models are presented in Table 7.3.[7] As expected, the measure of US financial system exposure is positively signed and statistically significant in each model.[8] Thus, jurisdictions most likely to receive a swap line from the Fed were those where systemi- cally important US banks and money market funds were more exposed. No other covariates appear to explain variation in swap line selection. For instance, although the coefficient on the share of trade with the United

Figure 7.9

US Financial Exposure and Swap Line Predicted Probability

States is statistically significant, it is negatively signed. Thus, increased trade with the United States is associated with a diminished probability of receiving a swap line. It is also worth noting that a shortage of dollar liquidity is not associated with an increased likelihood of a swap line when US financial exposure is accounted for.

To further illuminate the magnitude of the effect of financial exposure on swap line selection, Figure 7.9 displays simulated Model 2 predicted probabilities of receiving a swap line (with 90 percent confidence intervals) as financial system exposure increases from 0 percent to 2.5 percent. Although this may seem like a limited range, a 1 percent increase in this measure accounts for roughly an additional $50 billion in foreign exposures. As the figure indicates, the effect of financial exposure on the probability of a swap line is substantial. For countries where the US financial system was exposed at less than 0.5 percent, the probability of receiving a swap line is essentially 0. However, at 1.5 percent, the probability of receiving a swap line is nearly 1. This comports with a simple survey of the data. Of the 14 jurisdictions where financial exposure was highest, 13 received liquidity lines. Only India (1.1 percent exposure) was passed over.[9] Conversely, only one country where US exposure was below 0.5 percent—New Zealand (0.21 percent)—received a swap line. In short, consistent with my argument, the models show that the Fed was most likely to act as an ILLR on behalf of partners where the US financial system was exposed to significant counterparty risk.

  • [1] This decision is in line with Mahoney and Goertz’s (2004) "possibility principle,”which, simply put, states that when selecting negative cases, researchers should excludecases where—based on either theory or evidence—a positive value on the outcome of interest does not seem possible.
  • [2] Craig and von Peter 2010, p. 22.
  • [3] Once again, SIFI claims data were gathered from the CELS. These data are available athttp://www.ffiec.gov/e16.htm. Data on US foreign debt securities holdings were gatheredfrom the US Treasury’s Annual Cross Border Portfolio Holdings report available at http://www.treasury.gov/resource-center/data-chart-center/tic/Pages/fpis.aspx.
  • [4] Broz 2015; CRS 2009, p. 49; GAO 2011, p. 118.
  • [5] Data were obtained by the author from Allen and Moessner 2010.
  • [6] Allen and Moessner 2010.
  • [7] All models are fitted by using the R package Zelig (Imai, King, and Lau 2007, 2008).
  • [8] These results are consistent with Aizenman (2009) and Broz (2015), who each find USbank exposure correlates with swap line selection.
  • [9] A 2012 Bloomberg report sheds some light on why India was not selected for a swapline in 2008. Four years later, in November 2012, India reportedly asked the Fed to open
 
Source
< Prev   CONTENTS   Source   Next >