Today’s objectives

 

  1. Distinguish: policy analysis vs. basic research, with example
  2. Introduce: basic strategy evaluation framework
  3. Apply: use this framework to analyze U.S. decision to provide multiple launch rocket systems to Ukraine

Policy Analysis vs. Academic Research

 

 

 

There are 2 types of readings on the syllabus:

  1. Policy analysis
    • e.g. reports from researchers at RAND and other think tanks
  2. Basic (academic) research
    • e.g. journal articles by economists and political scientists

 

These two types of research products are complementary and mutually reinforcing.

 

But they are written for very different audiences, with different goals in mind.


Policy analysis

Basic research

Policy Analysis Basic Research
Primary audience Practitioners Scientists
Secondary audience General public General public
Main purpose Inform decision-makers Advance knowledge
Originating institutions Government agencies Universities
Think tanks Research labs
Common methods Program evaluation Experiments
Systems analysis Formal theory
Operations research Regression analysis
Dissemination Memoranda Journal articles
White papers Books
Op-Eds Patents
Roundtables Data and software
Metrics of impact Implementation as policy Scholarly citations
Grants Grants

These two worlds are not hermetically sealed, and there is cross-over.

Example: Counterinsurgency in North Caucasus, 2000-2012

Background: Second Chechen War

  • Chechnya becomes de facto independent after 1996
  • Chechen leadership split between moderates, Salafi-Jihadis
  • Salafi faction raids Dagestan in 1999, captures several villages
  • Russia invades Chechnya, captures Grozny and other large towns
  • insurgency begins, spreads to neighboring republics
  • Russia has great difficulty suppressing insurgency, violence lasts 10\(+\) years

Question:

  • which government strategies were most (or least) effective in reducing violence?


Insurgents

Security Forces

Strategic Options: how Russians respond to insurgent attacks

  1. Punishment: offensive operations to increase costs of fighting

    • detain or kill suspected insurgents in towns experiencing violence
    • tactics: artillery shelling, air strikes, sweeps, raids

  2. Denial: cordon operations to reduce insurgent mobility

    • physically isolate insurgents to restrict movement between towns
    • tactics: seal points of entry/exit, checkpoints, roadblocks, sieges

  3. Punishment + Denial: do both (e.g. cordon and search)

  4. No action: do neither

 

Russians relied on Option 1 in 78% of cases.

 

In one piece of basic research, Monica Toft and I asked “what if” the Russians had tried another approach (Journal of Peace Research, 2012).

Data: we parsed, geocoded incident reports from Memorial NGO

Insurgent Violence


Government Violence

Violence over Time

Research Design

  • we analyzed dynamics of violence using models from epidemiology

  • using village-level data, we modeled probability that rebel violence:

    1. spreads between neighboring towns (transmission)
    2. persists vs. subsides within each town (recovery)

    conditional on road network, government tactics, population density, unemployment, terrain, distance to military bases and admin centers

  • measure of effectiveness: basic reproduction number \(R_0\) \[\begin{align*} R_0=\frac{\text{transmission rate}}{\text{recovery rate}} \end{align*}\] if \(R_0<1\), insurgency will stop, not spread to new locations

    if \(R_0>1\), insurgency will persist and spread to new locations

  • we used simulations to see how \(R_0\) might change under different hypothetical government strategies (punishment, denial, etc.)

Simulation Design

  1. insurgents attempt to stage a series of attacks across region
    • 7,584 municipalities, connected by local road network
    • an attack can occur in any location with some probability
  2. where attacks occur, government responds
    • government consistently implements strategy \(k\)
      (\(k\in\) {punishment, denial, both, neither})
    • road network structure updated if/where denial is used
  3. insurgents strike again, government responds again
    • model predicts probability of new insurgent attack in each town, based on local conditions and historical patterns for strategy \(k\)
    • where attacks occur, government implements strategy \(k\) again
  4. repeat for 24 time periods (months)

We run simulation 100 times for each strategy, with different (random) starting locations for insurgent attacks.

Calculate average \(R_0\).

Which strategy is most (least) effective?

Simulation Results

Findings (lower values of \(R_0\) are better):

  • Denial is most effective at containing insurgent violence
  • Punishment is least effective, worse than doing nothing

 

What are some limitations with this type of analysis?

  • Assumptions about nature of Russian decision-making?
  • Assumptions about range of strategic options?
  • Assumptions about costs/resources/requirements?
  • Assumptions about time horizons?
  • Assumptions about how Russians evaluate success?
  • Other issues?

How might we structure this analysis differently if it was intended for a policy, not academic audience?

Comparing Multiple Policy Options

 

Strategy Evaluation in 3 Easy Steps

  1. Develop an analytical framework
  2. Perform the analysis
  3. Disseminate the findings

 

Application:

U.S. decision to send multiple launch rocket systems to Ukraine in 2022

Background: MLRS to Ukraine

  • Russia enters war with huge numerical advantage in tube artillery, MLRS
  • high rate of expenditure (>5,000/day)
  • no domestic production of shells, barrel replacements in Ukraine after invasion
  • Ukraine struggles to keep artillery units supplied, deliver counter-battery fire
  • by May 2022, Ukraine’s stockpile of 152mm Soviet-std shells nearly spent
  • U.S. considers sending 227mm M142 High Mobility Artillery Rocket System (HIMARS) to Ukraine
  • HIMARS has greater range, precision than other MLRS in Ukraine’s arsenal
  • but expensive, requires training, risks escalation due to high potential range


Destroyed BM-21

HIMARS

Step 1: Develop an Analytical Framework

  1. What is the problem?
    • define the trend/development/threat being addressed
    • state the level of urgency
    • estimate the degree of uncertainty
  2. What interests are affected by this problem?
    • define the stakes, whether they are vital/major/peripheral
  3. What goals will advance these interests?
    • state the desired “end state”
    • define intermediate objectives that need to be attained first
    • establish criteria for judging when each goal has been attained
  4. What policy options are available?

 

Application:

Let’s consider these Q’s in the case of whether to send MLRS to Ukraine

Step 2: Perform the Analysis

  1. What is the theory of cause-and-effect?
    • how the policy, if implemented, might help advance key goals
    • summarize existing evidence on whether this theory is valid
  2. What are the costs, compared to available alternatives?
    • breakdown: procurement, R&D, O&M, personnel, logistics, etc.
    • opportunity costs and trade-offs
  3. What is the implementation strategy?
    • division of responsibilities across depts/agencies/units
    • type and level of authorization required
  4. What are the time horizons?
    • speed of implementation
    • distribution of costs/benefits in short-term and long-term
  5. What is the feasibility of the policy?
    • flexibility and robustness to unanticipated events
    • degree of consistency/synergy with existing policies
    • likely sources of opposition / barriers to success
  6. What assumptions are being made?

How many of these questions can you answer in case of MLRS to Ukraine?

Step 3: Disseminate the Findings

  1. How wide should the distribution list be?
    • general public, including foreign
    • audience is within-government or within-agency
    • audience must have clearance, need-to-know
  2. What types of written products will be distributed?
    • memoranda (addressed to specific individuals)
    • white papers (in-house publications)
    • op-eds (accessible, for general public)
  3. What types of oral products will be prepared?
    • briefings (for specific officials, customers)
    • conferences/roundtables/talks (open to public)
    • interviews/media appearances
  4. How senior are the customers for these products?
    • different products needed for top officials, aides, staff
    • more senior \(\to\) shorter attention span

What type of dissemination strategy is most appropriate in our case?

NEXT MEETING

 

Economic Foundations: Land, Labor and Serfdom (Th, Sep. 12)

  • the “origin story” of Russian autocracy, imperial expansion
  • things to consider:
    • what incentives led Russia to adopt institution of serfdom
    • parallels and differences between forced labor practices in Russia vs. Western Europe vs. United States
    • why did the Russian state ultimately dismantle this institution?