A legislature capable of making laws broadly reflective of societal interests is a cornerstone for both development and democracy. Recognizing this fact, development agencies have invested millions in legislative strengthening initiatives around the developing world to develop legislative research capacities and provide legislators with information crucial to decision-making. At the same time, authoritarian regimes around the globe have adopted or revived institutions traditionally associated with democracy, primary among them the elected legislature. What remains unclear, however, is whether legislative strengthening initiatives in the authoritarian context induce greater responsiveness to societal interests, as intended. When properly informed of constituent preferences, to whom are authoritarian legislators more responsive - the citizenry or local firms? Is responsiveness improved at all?
Utilizing a randomized control trial (RCT), our project aims to establish whether targeted provision of constituent preferences increases the responsiveness of delegates to the Vietnamese National Assembly (VNA). In the first stage, we assign legislators to one of three groups: (1) those receiving infographics about the preferences of citizens within their province; (2) those receiving infographics about the preferences of local firms; and (3) a control group receiving no treatment. Because delegates caucus by provincial delegations prior to floor debates, the second stage of the RCT is a dose-response design in which we assign each province a treatment dosage, i.e. the fraction of delegates from a province receiving an informational treatment. Following the provincial caucus, we survey legislators about their views and the views of their constituents regarding the upcoming education law; delegate responsiveness to constituents is measured via dosage-interacted responses to a survey administered by the VNA library. We measure the responsiveness of delegates at the provincial level through mentions of the informational treatment during (1) internal, pre-debate provincial caucuses, and (2) VNA floor debates.
External Link(s)
Citation
Le, Anh et al. 2018. "Testing Legislator Responsiveness to Citizens and Firms in the Vietnamese National Assembly: A Field Experiment." AEA RCT Registry. June 20. https://doi.org/10.1257/rct.1608-3.0.
From delegate survey (at individual level, as interacted with dosage level):
Outcome 1: Whether a delegate has made up her mind.
Outcome 2: Whether a delegate cites lack of information regarding the preferences of citizens or businesses as the reason her mind is not yet made up. Lack of citation indicates greater responsiveness, and we will code it as such. Responsiveness=0 if delegate cites lack of information on constituency important in decision; Responsiveness=1 if delegate does not cite lack of constituency information.
From provincial caucus minutes (at provincial level):
Outcome 3: Whether any delegate (from a province) mentions the infographics, the statistics presented therein, or constituency opinions about educational quality.
Outcome 4: The number of times any delegate mentions the infographics, accompanying statistics, or constituency opinions about educational quality.
From floor debate transcripts (at provincial level):
Outcome 5: Whether any delegate (from a province) mentions the infographics, accompanying statistics, or constituency opinions about educational quality.
Outcome 6: The number of times any delegate mentions the infographics, accompanying statistics, or constituency opinions about educational quality.
Primary Outcomes (explanation)
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Experimental Design
Utilizing a randomized control trial (RCT), our project aims to establish whether targeted provision of constituent preferences increases the responsiveness of delegates to the Vietnamese National Assembly (VNA). In the first stage, we assign legislators to one of three groups: (1) those receiving infographics about the preferences of citizens within their province; (2) those receiving infographics about the preferences of local firms; and (3) a control group receiving no treatment. Because delegates caucus by provincial delegations prior to floor debates, the second stage of the RCT is a dose-response design in which we assign each province a treatment dosage, i.e. the fraction of delegates from a province receiving an informational treatment. Following the provincial caucus, we survey legislators about their views and the views of their constituents regarding the upcoming education law; delegate responsiveness to constituents is measured via dosage-interacted responses to a survey administered by the VNA library. We measure the responsiveness of delegates at the provincial level through mentions of the informational treatment during (1) internal, pre-debate provincial caucuses, and (2) VNA floor debates.
Experimental Design Details
Randomization Method
Randomization is conducted by an R script run at the investigators' home institution.
Randomization Unit
Provinces will be assigned to 0%, 50%, or 100% treatment on the basis of stratification variables (including GDPpc, fiscal transfers, delegation size, and summary statistics for individual-level variables). Within each province, delegates will be assigned to firm/citizen/control treatments on the basis of dichotomous stratification variables: nomination status, full-time status, electoral competitiveness.
Was the treatment clustered?
No
Sample size: planned number of clusters
0 clusters
Sample size: planned number of observations
471 delegates (490 active delegates minus 19 Politburo Standing Committee members)
Sample size (or number of clusters) by treatment arms
APPROXIMATE GOAL: 189 (40%) control, 141 (30%) firms' preferences, and 141 (30%) citizens' preferences.
ACTUAL SIZES: 181 control, 144 firms' preferences, and 146 citizens' preferences.
Differences due to the identities of those provinces (each with differing delegation sizes) assigned to each dosage level (0%, 50%, or 100%).
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)