第 32 個主題,共 64 個· 政府績效同交付指標

信任與合法性指標

合法性同支持,係 Mark Moore 喺《Creating Public Value》(1995)中「策略三角」嘅三條支柱之一——另外兩條係公共價值本身同營運能力——亦係最常冇被量度嘅一條,因為同預算或者產出數量唔同,合法性冇一個明顯嘅單一數字。信任同合法性指標係政府用嚟填補呢個缺口嘅一系列替代量度:機構信任調查、監督機構信心評級、投訴同上訴數據,以及政治/立法支持指標。

點解重要

Moore 嘅論點係,一個交付真正價值、卻失去政治同公眾合法性嘅公共管理者,最終會失去繼續交付所需嘅授權環境——資金被削減、授權範圍被收窄,服務無論成果幾好都會被餓死。因此,合法性唔係附加喺交付計分卡上嘅公關後記;佢係使命能否持續嘅承重輸入,所以先會喺公共價值計分卡中成為同等地位嘅視角,而唔係一個註腳。OECD 嘅「Trust in Government」調查計劃係量化呢個概念最領先嘅跨國嘗試:佢追蹤 OECD 成員國中表示信任本國政府嘅市民比例,長期數據顯示信任對衝擊高度敏感——2008 年金融危機同 COVID-19 疫情都引發全國層面嘅急劇波動,其後往往只係部分恢復,OECD 嘅分析一致發現,被認為嘅能力(政府有冇做到所講)同被認為嘅公平/誠信(政府係咪被視為冇貪腐或者偏袒),係信任數字嘅兩大最強驅動因素,有別於對任何單一交易嘅滿意程度。各國政府亦愈來愈嘗試喺更細緻嘅層面將合法性操作化——英國嘅獨立監管機構同視察機構(National Audit Office、Parliamentary and Health Service Ombudsman,以及 Ofsted 同 Care Quality Commission 等界別監管機構),擔當制度化嘅合法性檢查,將「公眾仲信唔信呢項服務」轉化為可審計嘅評級。

數學原理

信任同合法性屬於框架型課題,其可用嘅量化替代指標如下:

Institutional trust index (OECD-style)
  = % of survey respondents answering "yes" to a confidence-in-government question,
    tracked over time, disaggregated by demographic group

Legitimacy proxy set (no single number substitutes for the construct):
  - Upheld complaints per 1,000 service users (ombudsman or internal complaints data)
  - Judicial review / appeals success rate against the body's decisions
  - Independent regulator/inspectorate rating (e.g. "outstanding" to "inadequate" bands)
  - Legislative/oversight committee confidence votes or critical report frequency
  - Freedom of information request volume and disclosure/refusal rate, as a proxy
    for perceived transparency

Legitimacy is corroborated, not calculated: a defensible legitimacy assessment
triangulates several of the above rather than relying on any single proxy.

實例演算

國家稅務機關:為年度公共價值報告進行合法性三角驗證。

OECD-style trust proxy (department-specific confidence survey):
  58% of respondents say they trust the authority to "treat me fairly" (down from
  64% two years prior)

Complaints data:
  Upheld complaints: 4.2 per 1,000 taxpayer interactions (up from 3.1 per 1,000)

Ombudsman referrals:
  Referrals to the independent Adjudicator's Office: 1,850 in the year, of which
  61% upheld in full or part against the authority (up from 48% the prior year)

Reading across all three: trust is falling, upheld complaints are rising, and
independent ombudsman findings are increasingly siding against the authority —
three independent signals converging on the same direction, which is what makes
this a credible legitimacy finding rather than noise in any one series.

單一個數字變動只係薄弱證據;三個獨立量度喺同一時期一齊變動,先係令合法性判斷站得住腳嘅模式。

與軟件工程嘅關係

合法性指標好少由單一團隊嘅儀表板產生,而呢點本身就係設計啟示:應該建立可以攝取同核對獨立外部來源數據(申訴專員個案系統、監管機構評級饋送、問卷供應商)嘅報告管道,而唔係將合法性報告設計成純內部指標,因為內部來源嘅合法性聲稱(「我哋自評值得信任」)證據份量極低——同公共價值計分卡中合法性視角所指出嘅獨立性問題一樣。投訴同上訴數據管道,值得享有同任何為按成果付費合約提供數據嘅成果管道一樣嘅數據質素嚴謹度,因為一份漏報或者分類不佳嘅投訴數據集,會喺一年後信任調查中問題變得明顯之前,悄悄低估合法性問題。同呢項機構層面量度對應嘅交易層面量度,見市民滿意度指標;呢條支柱所屬嘅 Moore 完整策略三角框架,見公共價值。

常見陷阱

  • 將滿意度當作合法性嘅替代指標:市民可以滿意單一交易嘅界面,同時不信任整個機構(或者相反)——兩者必須分開報告嘅原因,見市民滿意度指標。
  • 倚賴單一自行匯報嘅指標:內部進行、冇獨立佐證(申訴專員數據、監管機構評級)嘅信任調查,好容易被視為自己打分而被忽略;應該三角驗證。
  • 忽略人口群組細分:全國總信任數字可以掩蓋特定群體(按年齡、種族、收入或地區)之間急劇分歧嘅合法性——OECD 自己嘅 Trust in Government 發布正正因為呢個原因作出細分。
  • 將單一衝擊導致嘅下跌讀作永久趨勢:信任數字喺危機(金融崩潰、疫情、轟動醜聞)前後急劇波動,並部分恢復;單一嘅衝擊後數據點,唔應該喺冇更多數據下推斷為長期下降。

資料來源