House of Lords - Economic Affairs

Select Committee on Economic Affairs Written Evidence

Memorandum by Dr Colin Rowat, Department of Economics, University of Birmingham

This submission comments on three issues related to economic sanctions: their effectiveness at achieving their stated aims; the effects of targeted and financial sanctions; and impediments to better targeting. It addresses these topics largely by appeal to the economics and political science literatures. Relative to other areas of economic policy debate it is immediately obvious that the economics literature is almost silent. Consequently, while considerable sanctions experience has been gained since 1990, spurring both a descriptive literature and institutional developments, policy is largely formed without the benefit of a rich analytical literature. This, in turn, reduces their ability to be precisely wielded.

1. ARE SANCTIONS "EFFECTIVE"?

The mordern analytical literature on sanctions' effectiveness dates to Hufbauer et al (1990). Published just after Iraq's invasion of Kuwait, and therefore on the verge of the "sanctions decade" (Cortright and Lopez, 2000), the authors identified 166 "major" sanctions episodes spanning the UK's imposition of sanctions on Germany in 1914 to the United Nations' on Iraq in 1990.[170]

The authors are interested in the effectiveness of economic sanctions as a policy instrument. Thus, to each episode they assign: a set of officially stated aims[171]; a measure of success (on a discrete scale from 1 to 4) in those aims; and a measure of the extent to which the sanctions themselves contributed to meeting those aims (again, on a discrete four point scale). An overall measure of sanctions' effectiveness is formed by multiplying their two four point scales, yielding a 16 point scale. An episode scoring nine or higher is judged a success.

The analysis are summarised in Table 1. The most widely cited figure from their analysis is the claim that sanctions have only worked in 34 per cent of cases.[172]

Table 1

SUCCESS BY TYPE OF POLICY GOAL


Policy goal	Success cases	Failure cases	Success ratio (%)

Modest policy change	17	34	33
Destabilisation	11	10	52
Disruption of military adventures	6	12	33
Military impairment	2	8	20
Other major policy changes	5	15	25
All cases[173]	41	79	34

Source: Hufbauer et al (1990, Table 5.1)

The authors also condition outcomes on various political and economic factors on episodes' success, one factor at a time. From these bivariate analyses, they derive seven[174] lessons for sanctions use:

175

2. "in most instances, multilateral sanctions are not associated with success". This follows directly from the limitations of their bivariate analysis: senders are more likely to build multilateral coalitions when their policy aims are ambitious.

3. politically or economically weak countries are more vulnerable to economic sanctions.

176

5. there is "a clear association ... between the duration of sanctions and the waning prospects of success", reflecting: a lost of appetite in the sender countries; mobilisation of resources in the target country; and the tendency of sanctions episodes between "allies" to be quickly and successfully resolved.

6. "sanctions that bite are sanctions that work", unless they attract a "black knight", a powerful third party who stands to gain from their failure.

7. "The more it costs a sender country to impose sanctions, the less likely it is that the sanctions wll succeed." In most "instances, the cost to the sender country in successful episodes is insignificant, and often the short-term result is a net gain (usually where the sanction is in the form of a cutoff of aid)."

While foundational, the Hufbauer et al (1990) conclusions may be best seen as suggestive due to a number of basic methodological weaknesses. Three that seem particularly significant are:

1. there is little formal discussion of their coding formulae, and no mention made of attempts to ensure inter-coder consistency. Thus, other researchers may reach different results.

2. "bivariate" analysis suffers from omitted variable bias and disallows that partialing out of multiple causal factors.

3. there is no theoretical framework; specifically, sanctions episodes are regarded as simply occurring, rather than resulting from the behaviour of agents with preferences.

The subsequent literature has tried to address some of these concerns.[177]

Pape (1997), for example, re-codes the Hufbauer et al (1990) episodes, concluding that:

An examination of the 40 cases in which HSE claim economic sanctions were successful reveals only five clear successes. The remainder are accounted for by 4 classes of errors: 18 were determined by force, not economic sanctions; 8 are failures, in which the target state never conceded to the coercer's demands; 6 are trade disputes, not instances of economic sanctions; and 3 are indeterminate. ... sanctions have been successful less than 5 per cent of the time.

Of the five "clear successes", "[t]hree of these were over trivial issues", such as embassy locations and the release of small numbers of prisoners. Even making favourable assumptions, Pape finds "no statistically significant relationship between target state GNP loss and sanctions success" in what appears to be a bivariate OLS regression.

In a subsequent exchange (Baldwin and Pape, 1998), Baldwin criticises Pape (1997) for using a binary standard of success, a coarser measure than the original four point scale; and for neglecting the costs of using economic sanctions.

Theoretically, Eaton and Engers (1999) claim that, when sanctions are the result of rational behaviour,

episodes in which senders actually take measures against targets provide a narrow window on the role of sanctions. In a world of perfect information, senders never actually resort to punishing. If the measure works, the target shapes up in anticipation. If not, the sender never threatens it initially. Hence, the very imposition of measures means that something went awry ... [I]ncomplete information (about senders or targets) can generate episodes in which measures are actually imposed, but these episodes constitute a very unrepresentative tip of an iceberg.

More specifically, if the target's resilience is not known to a sender, sanctions may only be imposed on targets which, deeming themselves insensitive to sanctions, refuse to comply when threatened. In these cases, simply observing actual episodes might lead one to "conclude that sanctions are futile".

On the other hand, if the sender's resolve is unknown but mutable over time, imposing sanctions may signal the sender's toughness, ushering in "an era of hegemony", and lead one to "conclude that punishment is extremely effective."

with more than one period of interaction, "[a] variety of outcomes can emerge" in which "[s]uccess is usually uncertain".[178]

Dresner (2003) explores the possibility of downward bias. Noting that information on sanctions threatened as well as those actually imposed can be hard to find (only five of the 116 Hufbauer et al (1990) cases end at the threat stage), he works with US datasets on trade policy, core labour standards and environmental regulations.

Table 2 displays the results. Rows display the number of episodes in which sanctions were merely threatened, as well as those in which they were actually imposed. In all three policy areas, sanctions merely threatened are more successful than those actually imposed. In the first area, this just fails to be significant at the 10 per cent level, while the difference is significant at the one per cent level in the remaining two areas.

Table 2

THREATS AND SANCTIONS IN US POLICY


	Nominal concessions	Significant concessions	Total	Success rate %

Trade policy
Sanctions threatened	31	40	71	56.34
Sanctions imposed	10	5	15	33.33
Core labour standards
Sanctions threatened	11	15	26	57.69
Sanctions imposed	7	0	7	0
Environmental regulation
Sanctions threatened	3	35	38	92.11
Sanctions imposed	18	20	38	52.63

2. TARGETED AND FINANCIAL SANCTIONS

Interest in targeted, rather than comprehensive, sanctions was given impetus in the 1990s by concern at the human cost of the comprehensive sanctions on Iraq.[179] Interest in targeted financial sanctions, in turn, has been given particular impetus after the 11 September attacks.

This recent interest means that the already small analytical literature on economic sanctions has even less to say about targeted and financial sanctions. Cortright and Lopez [2002a] apply a "back of the envelope" version of the Hufbauer et al. [1990] approach to 14 cases of UN sanctions applied over 1990-2001. They classify four as general or comprehensive, three of which they score as at least partially successful. Of the remaining 10 targeted or selective cases, they score two as successes, and all five of the pure arms embargoes as failures. They conclude that "the effeciveness of UN financial sanctions is uncertain. To date there is no case in which assets freezes and financial restrictions have been imposed on their own."[180]

Insight from the EU sanctions against the Federal Republic of Yugoslavia from 1998-2000, the US sanctions against Iran and the UN sanctions against Iraq are now presented. The FRY sanctions represent an example of careful efforts at targeting; the Iran sanctions have yielded a variety of cost estimates; the Iraq sanctions saw very slow reform of comprehensive sanctions in the direction of targeting.

The EU's sanctions' coordinator during the late 1990s FRY sanctiosn is cautious in his assessment of their success [de Vries, 2002]. His remarks recognise a lack of evidence about very basic aspects of the targeted sanctions:

A basic requirement for judging the effectiveness of sanctions is whether such measures deny targets the fulfillment of what they may consider an important or essential need. Evidence to that effect in this case is not available. Also unknown is the extent to which the targeted individuals were able to and actually did use false passports .... The freezing of funds may have been more effective in pressuring the regime although here again one should guard against strong conclusions. ...It was claimed that the financial sanctions had a major impact on trade between EU member states and the Federal Republic of Yugoslavia, but the evidence in support of this claim is ambiguous.

de Vries also describes Yugoslav authorities' frequent use of their power to redirect measures targeted at them onto the population more broadly: the oil embargo created rents which the elite were in the best position to capture; Yugoslave authorities threatened retaliation against EU airlines in response to the initial targeted flight bans; "white" companies exempted by the EU from asset freezes were "hit by retaliatory measures by the FRY regime"; these firms were also indirectly harmed through their economic links to targeted firms; cities receiving EU fuel under the "Energy for Democracy" initiative lost their regular state supplies; foreign currency could not practically be transferred to non-regime individuals given the regime's control of the banking system.

Table 3 shows annual costs of US sanctions to Iran around the year 2000 as estimated by four different analyists. That presented in Torbat [2005] crudely estimates trade losses based on pre-sanctions bilateral trade between Iran and the US, and financial losses based on the political risk premium paid by Iran. It and the Askari et al. [2001] study both find the financial sanctions to be more damaging than the trade sanctions; Preeg [1999] is ambivalent, and an Institute for International Economics case study finds the trade sanctions more harmful (in part by considering fewer aspects of the financial sanctions).

That financial sanctions may be more harmful than trade sanctions is consistent with Iraqi evidence. While the sanctions imposed on Iraq following the invasion of Kuwait, are frequently billed as the most comprehensive in the modern era they seem to impose very few barriers to trade, even as early as 1991.

Based on research conducted in Iraq in the autumn of 1991, Dréze and Gazdar [1992] noted that:

the present price of wheat flour is in fact close to the price that prevails in Jordan, which is Iraq's main trading partner. This "arbitrage" suggests that quantity constraints on import (and hence supply) are no longer binding. Comparing market prices of other staple goods between Jordan and Iraq, we also find almost complete arbitrage at the unofficial exchange rate.

In spite of various reform proposals over the years,[181] Iraq's borders remained under the control of it and its neighbours.

In both the Iranian and Iraqi cases, denial of capital market access may have been particularly harmful given their previous war damage.

After the passage of Security Council Resolution 1409 (May 2002), the Iraqi sanctions were advertised by American and British authorities as "smart" sanctions. However the careful thought about how to target individual regime members, and how to increase or relax pressure in response to regime behaviour that typified the EU approach to FRY was absent in the Iraqi case: SCR 1409 may largely have served to streamline administrative procedures for exporting goods to Iraq [Rowat, 2002]; it did nothing to help finance them, thus failing to ease one of the sanctions; most harmful constraints [Rowat, 2001]. Finally, the ambiguity dating back to April 1991 regarding what steps Iraqi authorities should take to lift the sanctions remained unaddressed.

Table 3

COSTS OF US SANCTIONS ON IRAN (ANNUAL US$ MILLION)


	Preeg [1999] (judgment)	Askari et al. [2001] (gravity model, judgement)	Institute for International Economics (welfare loss)	Torbat [2005] (welfare loss; price leadership; judgment)

Trade sanctions' costs	700-1,300	27	500	140
Financial sanctions' costs	800-1,300	1,160-1,321	250	637
Total sanctions' costs	1,500-2,600	1,187-1,348	750	777

3. IMPEDIMENTS TO IMPROVEMNT

Crafting well targeted sanctions policies requires both the existence of the institutional capacity to do so, and the political will to use those institutions.

In minor disputes between "erstwhile friends and close trading partners", these conditions are likely to be satisfied. When disputes are large relative to the perceived benefits of a relationship, less emphasis will be place on minimising harm to third parties in a target country. This section mentions two resulting political impediments to reducing harm to civilians.

First, the Security Council is unusually free of legal constrains, not explicitly accountable by the UN Charter either to other organs or to treaty law.[182] Perhaps its most important attempt to professionalise its sanctions administration, design, and implementation, under the Chowdhury working group (2000-01) failed even to issue a final report in the face of disagrerements among the Permanent Five.[183]

Second, tailoring sanctions to reduce civilian harm requires the target government's cooperation. The effects of any significant sanctions are difficult to assess without access to good data collected in the target country. The target government may have strong incentives to hide these data. Further, it has an incentive only to release data, or allow international organisations to release their own research, when it is in its interests; this, in turn, gives opponents grounds for dismissing any reports as politicised.

Additionally, many measures must be agreed by both the sender countries and the target country. The conflict evidenced by the imposition of sanctions reduces the set of feasible measures: sender governments may resist proposing measures that they expect to benefit the target government; the latter, in turn, may reject those felt not to benefit it.[184]

September 2006

170 This last episode is omitted from the authors' analysis as its outcome was not known at the time. An updated third edition has been forthcoming for some years, but has yet to emerge. Back

171 This authors are aware of a weakness of this approach, nothing that "Sender countries do not always announce their goals with clarity. Indeed, obfuscation is the rule in destabalisation cases". [Hufbauer et al, 1990, p 39] Back

172 Two of the three authoris presented preliminary extensions of the results above to the 1990s [Elliott and Hufbauer, 1999], when 50 new cases were launched. Sanctions in the 1990s seemed "about as successful | as the crop dating from the 1970's and 1980's". Back

173 Five cases are classified under two different policy goals | Since all but one of these cases are failures, double-counting them adds a small negative bias to the success ratio." Back

174 We omit the rather obvious "lessons" that: military policies are more apt to be used alongside sanctions when military aims are pursued; a sanctions policy should be carefully designed before being deployed. Back

175 This is especially likely to be the case in dictatorships. Back

176 Statistical significance is not considered. Back

177 See Hovi et al [2005], which refers to Hufbuaer et al [1990] as giving a "somewhat more positive" view of sanctions' efficacy, for mention of some of this literature. Back

178 Eaton and Engers [1992] address the repeated environment in greater depth, albeit under complete information. Back

179 Unicef's 1999 child mortality study was particularly influential, estimating on the basis of a 40,000 household survey that "if the substantial reduction in child mortality throughout Iraq during the 1980s had continued through the 1990s, there would have been half a million fewer deaths of children under five in the country as a whole during the eight year period 1991-1998." Back

180 The subsequent chapter in Cortright and Lopez [2002b] notes that the 116 cases in Hufbuaer et al [1990] only contain 13 cases of assets freezes. Elliott begins her own chapter in the same volume by claiming that: "With the exception of Libya, the results of UN targeted sanctions have been disappointing." Back