Currently there is a question of whether and how Afghanistan should engage non-state councils of elders to resolve disputes. In order to harness their benefits (e.g. efficiency), control their abuses (e.g. occasional controversial resolutions), and stabilize the resolution of disputes, some have argued that formal links should be established between the state judicial system and non-state councils of elders in Afghanistan. Others argue that mere informal links between the two systems should be established. Still others suggest that prior models of state engagement in Afghanistan should be revived. In Kyrgyzstan, just 65 miles north of Afghanistan, the Tsarist, Soviet and post-Soviet administrations have used various methods to engage non-state councils of elders. Despite these vastly different attempts at controlling local dispute resolutions, elders independently resolved disputes outside of Tsarist and Soviet control and continue to resolve disputes outside of post-Soviet state control in Kyrgyzstan. This phenomenon highlights the need for legitimacy in any models of state engagement that are considered for Afghanistan. If, as observed in Kyrgyzstan, people do not use state-sanctioned local councils in Afghanistan, then their purposes may be frustrated. Therefore, in addition to reviewing the experience with state engagement in other parts of the world, policy makers in Afghanistan should also review the historical and contemporary experience with state engagement in Kyrgyzstan. As the need for legitimacy is reconsidered and applied to the design of models for Afghanistan, it may be more likely that those models will be used by more people in Afghanistan and that their purposes will be furthered.

Several recent powerful tornadoes have raised interest among the public and policymakers in tornado shelters. We estimate casualty functions (fatalities, injuries) based on factors like storm intensity, county population density, and time of day, using historical data from Oklahoma and then estimate expected tornado casualties for Oklahoma using historical tornado frequencies assuming no further market penetration of shelters. We then calculate the cost per life saved and injury prevented for shelters. The cost per life saved for shelters in permanent homes without discounting is around $30 million. Shelters will then provide cost effective protection for residents of mobile homes. (JEL Q27)

Over the past several decades, engineers have made significant progress in the design and construction of structures able to withstand tornadic winds and debris. The aftermath of the 3 May 1999 F5 tornado in Moore, Oklahoma, highlighted the modest market penetration of tornado shelters in metropolitan areas. The authors use historical data from Oklahoma to estimate the potential casualties that tornado shelters could prevent and calculate that the cost per fatality avoided in single-family homes is $29 million while the cost per fatality avoided for mobile homes is $2.6 million. The estimates are sensitive to the proportion of strong (F3 or stronger) tornadoes and the choice of an interest rate for present-value calculations. If the F-scale distribution of Oklahoma tornadoes resembled a reported national frequency distribution and fatalities per category storm are held constant, the permanent home cost per fatality avoided triples to $88 million.

We examined the effectiveness of state automobile safety inspections and present new evidence from a panel of the 50 states for the years 1981-1993. Our approach incorporated several innovations over previous studies of safety inspections; most significantly, we estimated a fixed-effects model that incorporated state-specific shifts in casualty rates. We found no evidence that inspections significantly reduce fatality or injury rates. Our study also provides evidence on the effects of speed limits, seat belts, and Peltzman's offsetting behavior hypothesis.